The IPCC: dissolve it or not?

A couple of people have asked me this - I think it came up in Ask Stoat (I haven't forgotten, you know, just busy). Anyway, it seems like a great post - bound to be flamebait and get my comment count soaring!

You won't be too shocked to learn that I think it should be reformed, not dissolved. But how?

[Update: some of this gets quoted in http://www.guardian.co.uk/environment/2010/feb/10/ipcc-reform -W]

First of all, I think the WG I, II and II should be separated more. The WG I report should be done first. Then WG II and III should have their own timescales - perhaps running about a year or two behind WG I - and this makes sense because WG II and III need to use the output from WG I. It has been obvious to everyone for years that this would be sensible. WG II and III folk don't like it because they wouldn't get to ride on the WG I coattails the way they can now. Oh, and ditch the synthesis report, which is useless.

So why hasn't this been done already? That brings in the second bit of reform: political considerations sometimes overriding the science. Having WG II come out at the same time as I is clear evidence that the Wrong People are in charge of the timetable: that should be fixed. To a lesser extent this problem exists with the WG I SPM, too. But I think it is inevitable that there will be enormous and pointless fighting over the exact wording. And to some extent, desireable. The science is done by the scientists. The SPM headlines, that the politicians are going to have to act on, will have some political spin (and before the septics run wild, let me add that the spin so far has been always in the toning-things-down direction). Better written just by scientists, but too hard to manage to be worth wasting much time arguing about.

Make the whole thing less heavyweight and more selective. At the moment, getting your model run into the IPCC report is a badge of having Made It, which is why something like a quarter of the models in the AR4 are a bit crap, and some are awful. Everyone knows which are the good ones, though :-). And if we're being honest, we don't need that many modelling centres either. Too many modelling schedules are then driven by getting into the next report. Too many peoples papers are aimed at being the right degree of sexiness to fit in. And just too much good scientist time goes into preparation of the report. Perhaps they should just say less, and leave more in the papers. Stop repeating yourself: stuff that was basically right in the AR4 doesn't need to be said again in the AR5.

Stir up the WG II (and III?) people somewhat. While some of WG II is fine, it is clear that some sections have been edited by people who should not have been trusted by the job. It looks to me like some third world folk have been thrown their geographical section as a sop; it should be done more on merit. At the very least, get someone competent to review the edit comments for their sections.

[In support of the "some models are better than others" bit, I've found my connolley_bracegirdle_07_AR4_Ant_assess.pdf -W]

Refs

* 2014: the debate continues

More like this

I'm in the reform the IPCC camp. Stoat's suggestions seem like good ideas. Editing of the WGII needs some tightening up, and staggering the publication dates of WGI and WGII would allow for better results.

Disbanding of the IPCC goes too far, especially if there is no replacement. For some disbanding or over-reforming the IPCC are means to the end of sabotaging the science. That is a bigger concern than restoring the IPCC's public image.

By Joseph O'Sullivan (not verified) on 08 Feb 2010 #permalink

I say Dissolve IPCC, WG I,II, III as well as every other acronym, that isn't defined immediately before its usage.

By William Paysinger (not verified) on 08 Feb 2010 #permalink

WP

Oh dear, I just made an acronym out of William Paysinger without defining it first. I guess we should dissolve that too. ;o)

The IPCC seems to have become as laughable & meaningless as other "great" UN orgs such as the Commission on Human Rights (which has such "esteemed" member nations as Sudan and Zimbabwe, those bastions of human rights)! Let's face, the IPCC (fairly or not) is now about as much a laughing-stock as that "Commission on Human Rights" (which seems to be put together just to vent over Israel).

I feel that if the science is good, and most of it is, it can exist on it's own and in national bodies such as NSF, NERC, MetOffice, etc. But it seems looking back over the years, as soon as you ad the typical UN or EU-style political bullshit, you get stuff like WWF press releases being put in over actual peer-reviewed science etc.

[You seem to have fallen for the septic PR. No-one has found anything wrong with WG I, which is pretty good going for such a hefty tome, and even the WG II problems are small beer. There were no WWF press releases involved -W]

Maybe it's just the yank in my catching back up, but everything the UN touches seems to turn to mud...

1. Timing of WGI, II and III reports - done, but not by as much as you'd like. By the way, WGII is not as dependent on WGI as you'd like to think. If it was a forecast, yes, but it's not.
2. Developing country regional chapters - who do think should do these? US, UK, Aus, Canada, Germany etc? The AIACC research program (24 projects world-wide) funded 2001-2004 blooded many authors who ended up as WGII lead authors. These types of projects (which are very rare) can only improve global science, and we need more of it. The review editor process needs to be tightened up as you say.
3. Repetition - aside from short reviews of the previous assessment and advances, new material was the brief last time.

My suggestion after the AR4 was to create one report for adaptation and mitigation at the global level, and another report of ad-mit at the regional level.

The one critical piece of thinking that needs to come into the IPCC report process is a much better understanding of risk across all working groups. That will frame both the science and its application much better (e.g., it's not a forecasting exercise). Risk provides a framework for objective information and its application in normative situations.

By Roger Jones (not verified) on 08 Feb 2010 #permalink

I think that IPCC should finish the current rount (AR5) anyway. But there may still be a choice between a relatively small number of people focusing the global changes and a relatively large number of people dealing with both global and regional climate change.

After AR5, I think there should not be IPCC AR6. Rather I hope a drastic reorganization.

One example of the idea, very UN-centric, of the reorganization is as follows. It will be the UN Panel for Sustainable Environment and Society. Its WG1 will follow up IPCC (WG1+WG2) and other issues of physical environment (e.g. ozone loss, acid precipitation). WG2 will follow up the Millenium Ecosystem Assessment as well as scientific assessments for UN Convention on Biological Diversity. WG3 will follow up the social-scientific assessments for Millenium Development Goals. WG4 will formulate policy options to accomodate issues of other three WGs.

Dissolving the IPCC? It must be a joke... It (at least the WG I) is the best compendium of science for the general public ever. Not being a scientist, it's been my only chance to fully satisfy my scientific curiosity in depth. I wish other scientific fields had the same assessments. A shorter update might be a good idea, but I think you will bump into the same problems as in the SPM or the Synthesis Report: how to communicate the uncertainties.

Everyone knows which are the good [models], though

I don't have a clue. In your opinion, which are the best models? (thx)

[Ah, well, I'd probably put MPI / Hadley at the top, but then I would, wouldn't I? Stuff like Connolley, W. M., and T. J. Bracegirdle (2007), An Antarctic assessment of IPCC AR4 coupled models, Geophys. Res. Lett., 34, L22505, doi:10.1029/2007GL031648. can help you. I wouldn't want to be too indescrete here, and anyway I've forgotten the details -W]

just to respond to your comments above -- as you say it's become basically a status symbol for various modelling groups, and basically turned scientists into even worse politicians than the usual academic backstabbing & schadenfreude. and stuff like the "glaciers in 2035" is not "small beer" IMHO -- it just points to the political nature the IPCC has devolved into. so I think it's long past getting rid of it. it had a half-decent run and it seems to be repeating itself and needing to have more alarmist stuff in it to get the publicity. the latter which will just give more embarassing wrong info like the glaciers, Netherlands underwater, etc.

I come up with a suggestion to the current round of IPCC. Recruit "Reliable Source" checkers from the Wikipedia community!

I read "Talk" pages of Wikipedia. I read the news of a few failures in IPCC. Deja vu. Both institutions have very similar ethical codes: Need Reliable Sources. No Original Research. These seems unusual elsewhere.

I think that persons who are good at RS and OR checking are more likely to be found among Wikipedia editors than among professional scientists who have been trained mainly to write about their original research.

Of course, the checkers must also be able to understand the subtle difference between these communities, and adapt. So, not all Wikipedia editors are eligible.

[Ha, maybe. What about all the "climate auditors" who claimed to be so diligent? Where were they? Minutely poring over the wrong stuff, I think -W]

I think Kooiti is on to something. WG 1 needs to be done on a continuous reassessment basis, so a wikiPCC project might not be such a bad idea. For genuine IPCC contributors, of course, not rabble like us :-p

"Stuff like Connolley, W. M., and T. J. Bracegirdle (2007), An Antarctic assessment of IPCC AR4 coupled models, Geophys. Res. Lett., 34, L22505, doi:10.1029/2007GL031648"

I can't find an open online version. Are you able to post the pdf somewhere, or at least send me a copy if you're old-fashioned enough to adhere to publisher's copyright claims :-)

By Jonathan Baxter (not verified) on 09 Feb 2010 #permalink

WGII is supposed to express the peer-reviewed consensus on impacts, so publishing it a year after WGI discusses science means there's no intervening peer review to translate the science into impacts. I think it would make more sense to schedule WGII three or four years after WGI, so that peer review can consider the consensus on the science, translate it into impacts, and WGII can attempt to deliver the new consensus on impacts.

I think the delay is fine, and would remind people of the problem more often than one major IPCC publication every seven years.

"After AR5, I think there should not be IPCC AR6."

But Masuda-san, we already have people working on the model for AR6. What would they do?

:-)

Dr. Andrew A. Lacis - NASA GISS

comment on Executive Summary Chapter 9 AR4 [of, I think you mean, the WG I chapter 9 - WMC]

[The usual: {{cn}}. You found this somewhere; where. Septic blog? IPCC? Who knows. If you mean http://dotearth.blogs.nytimes.com/2010/02/09/does-an-old-climate-critiq… then that supplies your answer... that this is based on a misunderstanding of the meaning of the Exec Summary. All clear now? -W]

The presentation sounds like something put together by Greenpeace activists and their legal department. The points being made are made arbitrarily with legal sounding caveats without having established any foundation or basis in fact. The Executive Summary seems to be a political statement that is only designed to annoy greenhouse skeptics... [cut - WMC]"

Rejected. [Executive Summary] summarizes Ch 9, which is based on the peer reviewed literature.

By windansea (not verified) on 09 Feb 2010 #permalink

William,

If the objective is to persuade China and India to reduce their carbon emissions, you need to ask, will any amount of reform cause the Chinese and Indian governments to trust the IPCC brand again?

See e.g. http://www.energytribune.com/articles.cfm?aid=3071

Best regards,
Alex

[I think you have a very naive view of how things work. The fluff over 2035/2350/whatever is just fluff. At the moment, it is really exciting fluff, but in a years time it will be no more exciting that the satellite temperature record (remember how that conclusively disproved global warming, back in the old days, before it, err, didn't), or ocean cooling, or any one of a number of shallow stories.

The fluff provides an excuse for anyone who wants to do nothing, to do nothing. But they would have found an excuse anyway. The Chinese, and the Indians, know where the state of the science actually lies and are not mislead by this, any more than the Bush administration was mislead - despite their foot-dragging on climate change, they actually knew exactly where things stood - they had to, it was even more important to them, so they could lie effectively -W]

By Alex Harvey (not verified) on 09 Feb 2010 #permalink

"that this is based on a misunderstanding of the meaning of the Exec Summary. All clear now? -W]"

All cleared up by one "Gabriele Hegerl", who seems to believe in very high sensitivities.

Check out Figure 2. I've seen it peddled by a number of alarmists. Note the big scary fat tail on the y axis. That's what happens when you pass a Gaussian through a singular filter (1 / 1-f).

Unfortunately, no-one in the climate science community seems to realize that the temperature distribution you get via that transform (the y-axis in Figure 2) has infinite expectation, and hence is worthless. Or perhaps they do realize and it suits their alarmist agenda not to discuss it - who knows? (This is not some two-bit journal, it's Nature, so one presumes these kind of articles are reviewed by the top people in the field).

A reasonable person would conclude that the distribution is misparameterized, not that we're all going to fry (with probability 1!)

[You need to get yourself up to date. Head over to James Annan's blog and have a browse around through the thicket of pictures. http://julesandjames.blogspot.com/2006/09/can-we-believe-in-high-climate.html might do, there is more there -W]

By Jonathan Baxter (not verified) on 09 Feb 2010 #permalink

Not with probability 1. That was OTT. But the T distribution still has infinite expectation, which is nuts.

By Jonathan Baxter (not verified) on 09 Feb 2010 #permalink

You found this somewhere; where.

I looked at the most likely place, and there it was: WUWT.

Watts seems to miss the point that critiquing the structure and content of a presentation doesn't mean that the scientist being quoted doubts the underlying science. And in this case, the scientist in question most assuredly does not.

Phil: I did not mean to immense IPCC with Wikipedia culture. I just wanted to say that Wikipedia may be the place where IPCC should search for human resources.
But it may be a good idea for something which follow up IPCC to take more elements of Wiki culture. I think that its editorship need be more elitistic than Wikipedia, though.

James: I think that IPCC should "be dissolved developmentwise" (literal translation of Japanese "hatten-teki ni kaisyoo-suru"). I am not sure what it should develops into, but, I believe that the activities of IPCC WG1 will somehow continue (though emphasis on prognostic experiments global coupled climate models may attenuate). So let's prepare for AR1 of the new panel!

(I guess that someone who decided to call the First Assessment Report "FAR" did not expect that IPCC would continue so long as to its 4th Assessment Report. I see many occasions where "FAR" actually means AR4. I saw one on in Wikipedia and I corrected. It was the first occasion of editing Wikipedia.)

"[You need to get yourself up to date. Head over to James Annan's blog... - W]"

I have read James Annan's Bayes stuff. In fact a preprint before it was published in 2005 (from memory). The article I linked to was published in Nature in 2008, by what looks to be one of the leading players in the IPCC. The paper she refers to in Figure 2 (with the infinite expectation on T) was published in Science in 2007. If you read that paper it is clear the authors were not aware (and apparently are still not aware) that their putative T distribution does not possess a first moment.

Annan's stuff is an interesting application of Bayes' rule to a bunch of different -- nominally observation-based -- estimates. But you don't need Bayes to see the flaws in these bogus alarmist arguments based on wacky pdfs. (That said, I *would* reccomend Hegerl's paper as a good review article, just don't rely on the analysis too much).

[The reason I pointed you at JA is because he makes very similar criticisms, and has papers saying so. In fact, from a scientific POV these climate sensitivity issues are *far* more important than the wibbling over the 2035 date. The difference being that they are sufficiently complex that they don't make a good sound bite for the journos -W]

By Jonathan Baxter (not verified) on 10 Feb 2010 #permalink

"[In fact, from a scientific POV these climate sensitivity issues are *far* more important than the wibbling over the 2035 date.]"

Agreed. But wibbling over 2035 is good sport.

"[The difference being that they are sufficiently complex that they don't make a good sound bite for the journos -W]"

No? There have been numerous press releases and breathless stories along the lines of "climate sensitivity may be as high as XX" for some ridiculous XX. Accompanied by scary maps glowing luminescent red. Stern played that game. Hansen plays that game. Hegerl plays that game. It's everywhere, and widely reported.

[Of course. What I meant is that pointing out the error, and why, is far too difficult for sound bites. Plus even understanding the issue is tricky. Most skeptics can cope with the difference between 2035 and 2350, even if they have trouble with the IPCC rules :-). Understanding moments of functions is going to go straight over their heads -W]

By Jonathan Baxter (not verified) on 10 Feb 2010 #permalink

"Understanding moments of functions is going to go straight over their [skeptic's] heads -W"

I don't understand your point. It's not the skeptics that are the problem. It's the very famous experts. Hansen, Hegerl, Stern, realclimate. They're all guilty of not understanding that the fat tail on climate sensitivity is simply an artefact of how the distribution is parameterized.

[Oh dear. Look, let me do this very slowly. It is an error. But unlike trivial errors like 2035, it has not been widely publicised. This is because unlike that trivial error, it is not an error that is easy to "sell". This is how you can tell that the septics have no real interest in the science - they are only interested in knocking the IPCC. They are looking for interesting soundbites to sell; they are not interested in actually examining the science. Unlike scientists, such as JA, who are and have.

I'm also not convinced by your list of people who have got it wrong; to take one, what makes you think RC has got it wrong? -W]

Hegerl says explicitly (page 739):

"A further fundamental reason for the fat tail of S is that S is proportional to 1/(1 â f)

...

if the uncertainty in f is assumed to be Gaussian."

Not only does S have a fat tail, it has infinite mean. So reductio ad absurdum, f cannot be Gaussian. But instead of that conclusion (or concluding that the whole exercise of trying to model uncertainty in S by modeling uncertainty in f is suspect), the "experts" (with a few exceptions) trot out variations of the precautionary principle: "ooh look, we can't bound climate sensitivity above so we better decarbonize the entire economy today, just in case".

You can draw any conclusion you like with that kind of argument. Take your poorly understand, potentially harmful variable (eg, I dunno, gamma-ray flux through the stratosphere - call it S), relate it to some other poorly understood quantity f via f = 1 - 1/S, then claim f has a Gaussian distribution (don't they all?). Presto, S has a fat-tail, and we'd better do something about it real soon otherwise we're all going to die hideous and painful deaths.

By Jonathan Baxter (not verified) on 10 Feb 2010 #permalink

Kooiti said:

Phil: I did not mean to immense IPCC with Wikipedia culture. I just wanted to say that Wikipedia may be the place where IPCC should search for human resources.
But it may be a good idea for something which follow up IPCC to take more elements of Wiki culture. I think that its editorship need be more elitistic than Wikipedia, though.

Oh, I think the normal IPCC reviewers should edit it, but a wiki form with a running discussion could be useful. The nits would be picked in context.

But I still think we need an ongoing work in progress, online, continually updated, in place of the olde worlde concept of infrequently published reports.

"[This is how you can tell that the septics have no real interest in the science .. -W]

I am one of your so-called "septics". And I am not the "exception that proves the rule". Plenty of skeptics have been pointing out the error in the way climate scientists and policy wonks have been reasoning about uncertainty, although maybe not in this explicit form. Lindzen, Roy Spencer, McIntyre, Richard Tol, just off the top of my head. You're making the mistake of characterizing all skeptics by the lowest common denominator.

Furthermore, this isn't some subtle error. It is a basic failure of scientific reasoning.

"I'm also not convinced by your list of people who have got it wrong; to take one, what makes you think RC has got it wrong? -W"

Eg,this:
"(The tail in S results from the probability distribution of the feedback strengths, and unless those uncertainties are distributed very, very differently than the Gaussian distribution assumed by Roe and Baker, the tail will remain)."

Of course those uncertainties are "distributed very very differently than the Gaussian". For starters the distribution is truncated on the right by 1. And for values close to 1 the climate goes mad which we have zero evidence for, so as f approaches 1 from below the probability must go rapidly to zero which the Gaussian does not do.

Think about it another way: why do we use Gaussians? Because the central limit theorem tells us the distribution of repeated measurements of the same quantity approaches normal (pretty rapidly) as the number of measurements increases. But in the case of climate science we don't even have a *single* decent measurement of f (or S). So there's no justification for using a Gaussian.

By Jonathan Baxter (not verified) on 10 Feb 2010 #permalink

> we don't even have a *single* decent measurement of f (or S)

Do you have a proposal for making those two measurements?

Do you understand we know how doubling CO2 rapidly affects climate, looking at the paleo record?

There are any number of distributions that have infinite means, but meaningful medians, the St. Petersburg Paradox among them. Objections such as Jonathan Baxters are basically formalistic, requiring that one get unreasonably close to a pole, that is the models are valid in regions where they are used, and the divergence occurs in a place that is physically impossible to reach.

@Hank I understand Hansen has argued he knows how doubling CO2 will affect the climate based on paleo records, but he gives no justification for his implicitly linear assumption (He gets today's sensitivity by dividing the temperature change between now and the LGM by the forcing change over the same period (factoring out ice-albedo effects). That only works if you assume the sensitivity is constant. If it was, for example, much higher at the LGM -- not out of the question since the earth was a lot colder and drier -- then today's sensitivity may be much lower.)

@Eli The whole argument in Roe and Baker (and Realclimate and Hegerl and..) relies on getting "unnecessarily close to a pole". They all talk about the "fat tail". You're not in the tail if you are far from the pole.

By Jonathan Baxter (not verified) on 11 Feb 2010 #permalink

BTW, this is awesome

(US-East-Coast-Centric. We're suffering an all-time record snowpocalypse).

By Jonathan Baxter (not verified) on 11 Feb 2010 #permalink

Well, according to some of the commenters on the Grauniad's site, he's... Oh, let's not go there :-)

Hank, are you referring to this:

"Do you have a proposal for making those two measurements?"

I assumed you were being rhetorical. No-one seems to know how to measure sensitivity directly. The paleo approach is the closest, but suffers from the linearity question. Volcanos are potentially a good source, but judging by the arguments people get into we are still not measuring all the relevant variables.

Maybe we should try measurements in a box? After all, we test aircraft models in wind tunnels. Or maybe simpler models? Personally I think this obsession with whole-earth modelling is somewhat misguided. They're so complex it is difficult to know if the code is correct, and there are so many variables it is difficult to isolate their influence. I can't help thinking a carefully crafted ideal model might be easier to analyze and potentially yield more robust conclusions, but I am no expert.

By Jonathan Baxter (not verified) on 11 Feb 2010 #permalink

>"[In support of the "some models are better than others" bit, I've found my connolley_bracegirdle_07_AR4_Ant_assess.pdf -W]"

Extended article on [[ECHAM]] http://en.wikipedia.org/wiki/ECHAM

Now if you don't revert that, this will be clear evidence that leave what you like and delete what you don't like on the Wikipedia climate pages. ;o)

[Was any proof needed? :-) -W]

Doh!!! Please point me to a wikipedia editor other than a vandal that doesn't.