Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace

Discussion > Lukewarming

Feb 8, 2013 at 5:39 AM | shub

Blaming the avg temp increase in the second half of 20th century on CO2 is. It is a very specific statement and claim.

What is the lukewarmers stance toward this IPCC statement?

Not to be nit-picky or anything like that ... well, OK, I suppose I am ... but doesn't the IPCC statement (albeit in a very pea and thimble round-about way) "blame" most of this putative increase on human generated CO2?

If I'm not mistaken in my "interpretation" of the IPCC's (deliberately deceptive?) imprecision, I'd be interested in knowing what evidence the Lukewarmers might have in support of this very tenuous correlation.

Feb 8, 2013 at 8:22 AM | Registered CommenterHilary Ostrov

Let me try to clear a few things up. There seem to be two perceptions of what I said that were unintended, so I'd like to clarify:

1. Lukewamer is a position, not an attribution.

If you look back at my post a few days ago in this thread on Feb 6, 2013 at 11:16 AM

The scientific skeptic has to be open to either of these models [natural or CO2] being ultimately proved correct. Since both produce a small amount of warming in the next century, whichever one is right, leads to being a Lukewarmer.

People get to being Lukewarmer by different routes, and they may believe it will be lukewarm for different reasons. For a scientific skeptic weighing up the options presented, the only ones which are supported by the data (natural and small GH) lead inexorably to the belief that warming will be Lukewarm.

So I'm not claiming that Lukewamers have an attribution they all agree on, only an outcome they all agree on. You can work forward from the attribution to the Lukewarmer outcome, but you can't work backwards from the Lukewarmer position to a particular attribution.

2. The null hypothesis for the 'warming' question is not the same as the null hypotheses for the constituent possible explanations.

As someone pointed out earlier in the thread, the null hypothesis for "Can CO2 warm an atmosphere?" is "CO2 does NOT warm an atmosphere" - this is null hypothesis for THAT particular question.

By the same token, the null hypothesis for "natural fluctuation" question is "Natural fluctuation does NOT warm the atmosphere". The null hypothesis for the "aliens from Mars" question is "Aliens from Mars do NOT warm the atmosphere". Stating the null hypothesis for individual explanations is all well and good, but it doesn't get us anywhere, and doesn't get us to the actual question that I posed way back at the start:

"Given that warming has taken place (and we believe that record), what caused it, and what is the null hypothesis for this question?"

Given all the hypotheses (and ignoring the stupid ones through Occam's Razor):

"No change at all"
"Natural fluctuation"
"Interglacial rise"
"Increase in GHG"
"Positive feedbacks"
"Something else we don't know"

Which one is the null hypothesis? Is there a single one?

You can't say the null hypothesis is "No change at all" because the warming is assumed in the question, so the null hypothesis breaks down into the options:

1. None of those explanations explain it
2. One or more of those explanations explain it either exclusively or additively

Since we have to assume we've nailed the possibles, we have to discard 1, until someone comes up with something else.

So the null hypothesis for the "given warming, what caused it" is "one of those explanations caused it"

We can narrow it further by discarding hypothesis that don't fit the data:

1. No change can go because it's flatly invalidated by the data
2. Positive feedbacks can pretty much go, because they have not been observed, so invalidated
3. 'Something else' can go for now, until it is defined

This leaves us:

"Natural fluctuation"
"Interglacial rise"
"Increase in GHG"

Now each of these hypothesis have their own null hypothesis, and own level of data, proof and etc.
Also, each of these is NOT mutually exclusive - they are additive - they all affect a single metric - temperature.

We don't understand any of them well enough to say which one is more likely. By the null hypothesis it has to be one or more of them, and could be all of them.

This is as far as a rational skeptical scientist can go. There is some evidence for all of them. Do we have credible models for all of them? No. Are they all equally likely? Who knows. More data required.

What is the outcome in terms of termperature for each of the scenarios:

"Natural fluctuation" - no or small rise in the next century
"Interglacial rise" - small rise in the next century
"Increase in GHG" - no or small rise in the next century
Combination of all/any of the above - small rise in the next century

Lukewarm.

Now you understand my position, I'm not trying to convert anyone and bully them or silence disagreement. I'm just demonstrating where my own thinking has gone. Because it's based on invalidating hypothesis according to data, if new data comes in then the chain of logic changes direction.

Hope this clears it up.

Feb 8, 2013 at 9:20 AM | Unregistered CommenterTheBigYinJames

And the "natural variability" stuff is unscientific - what exactly is varying?

Whatever it is that has always been varying?
Feb 7, 2013 at 10:47 PM Martin A

and if you wanted to model that, Martin A, how would you do that?
Feb 8, 2013 at 12:01 AM diogenes

Buggered if I know.

And I'll be buggered if anyone knows - in view of the complexity of the system, the lack of understanding of its dynamics (and of its physics) and the total absence of detailed information about its past.

____________________________________________________________________________________

The whole of "climate science" seems replete with models which its practitioners seem to have trouble distinguishing from reality. They blithely talk about "experiments" they conduct using their unvalidated models. They use the output of one unvalidated model to drive another unvalidated model and appear to believe that the final result has some sort of validity.

Some appear to have some sort of grasp of the notion that a model needs to be validated before it is any use - but they then go and blow it by revealing that they don't understand that validating a model is something different from checking that it can reproduce the data used to construct it.

Feb 8, 2013 at 9:46 AM | Registered CommenterMartin A

There is no such thing as "the" Lukewarmer position. Indeed there are several different approaches to defining it, though it should be noted that many Lukewarmers fit many definitions. Broadly speaking, Lukewarmers believe that CO2 warms the planet, but not that much. The two halves of that belief usually have quite different origins.

Lukewarmers believe that CO2 warms the planet because simple physics says that greenhouse gases do that, and that CO2 is a greenhouse gas. Note that this argument is frequently taken far too far by proponents of the consensus: simple physics does not lead to any particular value of the climate sensitivity; that requires rather a lot of rather complex physics, the details of which are still poorly understood. But that CO2 warms the planet by some (as yet unspecified, and quite possibly small, or even negligible) amount is straightforward.

Lukewarmers understand enough physics to see how simple radiation arguments lead to warming. Of course they also understand enough physics to see that these naive models are hopelessly oversimplified, missing out all sorts of important elements. But crucially they understand enough physics to see that missing out all these complexities is not going to change the sign of the response.

So Lukewarmers believe that CO2 warms the planet. They may also believe that climate sensitivity, defined roughly speaking as dT/dlog[CO2], is a useful concept. Note that this does not require you to accept the whole approach of putting all influences on a common "forcing" basis, but only to accept this result for well-mixed greenhouse gases. Similarly it is perfectly possible to believe that the climate sensitivity is only a useful concept for small changes around a given climatic state, and that it might take different values over time. But most Lukewarmers would agree that CO2 induced warming will be approximately proportional to logarithm of concentration, and so climate sensitivity is a handy (if rough and ready) summary.

Since they accept the concept of climate sensitivity Lukewarmers will probably have some opinion as to its value. They will of course believe that it is greater than zero (though it could be only just greater than zero). However to be called Lukewarmers they pretty much have to believe that the sensitivity is lower than the IPCC thinks. This could stem from a disbelief in the long high sensitivity tail, but usually involves a belief that the best estimate is lower than the IPCC thinks.

This leads to my personal definition of Lukewarmer: somebody who (1) believes in basic radiative physics, (2) believes that climate sensitivity is a vaguely useful concept, and (3) believes that the climate sensitivity probably lies between 0 and 3K per doubling. As various people have noted this is not that different from the conventional AGW position, which would agree on (1) and (2), but place the sensitivity between 1.5 and 4.5 K per doubling. (I am tactfully ignoring the extreme CAGW position, which blathers on about the long tails and thinks we should take 9K per doubling as a serious possibility.)

Not everyone agrees with me on this. Steven Mosher seems to think that a Lukewarmer believes that the climate sensitivity lies between 1.2 and 3.5K per doubling. I really can't understand his arguments for this position, but given his historical importance in the Lukewarming movement it is a position which has to be taken seriously. However I do think it is fundamentally flawed: by his definitions Lindzen (and Spencer?) are not Lukewarmers, while by my definitions they both clearly are.

Why do Lukewarmers favour this range? Well as we are all good Bayesians we can trace this belief to either their priors or their likelihood functions.

Many Lukewarmers want to use noninformative priors, which pile up their density at the low end, while the consensus has a bizarre preference for uniform priors, even though almost everybody admits this is wrong. Other Lukewarmers start from personal "expert priors" which are usually peaked around the no-feedback sensitivity. This strikes me as a reasonable approach: the physics is well enough understood to calculate a no-feedback sensitivity with some sort of confidence, but the physics of feedbacks is sufficiently poorly understood that it is reasonable to treat these as a priori unknown, and so needing to be derived from experimental data, such as temperature measurements. However you do still need to have some argument to justify the width of the prior distribution around the no-feedback centre.

The likelihood arguments are more complex, but basically come down to a belief either (1) that the temperature data is partially corrupted, or (2) the likelihood function people are using contains wrong corrections for confounding factors such as aerosols.

Personally I have considerable sympathy for the Lukewarming position, but as I have said before this is not my field. I also understand Rhoda's position (a charming person, and a fine drinking companion). In a sense Rhoda, not being an academic, has no interest in the climate sensitivity: Rhoda just wants to know whether there is any direct evidence that any of this actually matters in the here and now. "Do not worry about tomorrow, for tomorrow will worry about itself" is a pragmatic position, and certainly not crazy.

Feb 8, 2013 at 11:17 AM | Registered CommenterJonathan Jones

Excellent reply. My prior is based around the non-existence of previous catastrophic amplification due to feedbacks in the paleo record. While not a quantatative prior, in quality it restricts the spread to a small range.

Feb 8, 2013 at 12:48 PM | Unregistered CommenterTheBigYinJames

shub

Thanks for your response, you certainly made an early start today.

I'm pleased to see I did not misunderstand your position regarding the null hypothesis for what is the crux of the whole climate debate ie. that increased atmospheric CO2 is the cause of increased global temperature.

With respect to the accepted catastrophist/warmist/lukewarmist positions, I'm looking forward to a few more responses to your question to gauge how the average person understands the lukewarmist stance.

Feb 8, 2013 at 12:57 PM | Registered CommenterRKS

"Lukewarmers believe that CO2 warms the planet because simple physics says that greenhouse gases do that, and that CO2 is a greenhouse gas. Note that this argument is frequently taken far too far by proponents of the consensus: simple physics does not lead to any particular value of the climate sensitivity; that requires rather a lot of rather complex physics, the details of which are still poorly understood. But that CO2 warms the planet by some (as yet unspecified, and quite possibly small, or even negligible) amount is straightforward.

Lukewarmers understand enough physics to see how simple radiation arguments lead to warming. Of course they also understand enough physics to see that these naive models are hopelessly oversimplified, missing out all sorts of important elements. But crucially they understand enough physics to see that missing out all these complexities is not going to change the sign of the response.

(...)

Since they accept the concept of climate sensitivity Lukewarmers will probably have some opinion as to its value. They will of course believe that it is greater than zero (though it could be only just greater than zero). However to be called Lukewarmers they pretty much have to believe that the sensitivity is lower than the IPCC thinks. This could stem from a disbelief in the long high sensitivity tail, but usually involves a belief that the best estimate is lower than the IPCC thinks."

Feb 8, 2013 at 11:17 AM Jonathan Jones


How delightfully clearly put. Thank you for your contribution.

In that case, I count myself as a lukewarmer - though I'd put my guess for the "sensitivity" as somewhere between 0.0...01 and 1.0, with a bias towards the lower end of my log scale of sensitivity.

__________________________

Diogenes said: " And the "natural variability" stuff is unscientific - what exactly is varying? "

To which I suggested: "Whatever it is that has always been varying?"
Feb 7, 2013 at 10:47 PM

Diogenes then asked: "and if you wanted to model that, Martin A, how would you do that?"
Feb 8, 2013 at 12:01 AM

I replied that neither I nor, so far as I can see, anyone else can model the natural (ie not caused by mankind) variations of the climate. But I don't think I made clear what are the implications of this.

If it is true that no-one can produce an accurate and validated model of the natural variations of climate, I think this amounts to saying "We don't understand the natural variations of climate".

If we don't understand the natural variations of the climate, what earth are "climate scientists" playing at when they claim they do understand the effect on climate of a change in a minor greenhouse gas?

[And without starting to discuss the lack of understanding of the dynamics of the atmospheric concentration of CO2, the claims of the authors of the "Bern model" not withstanding.]

To me it would be utterly ludicrous to say "Here is a supremely complicated dynamic system with all sorts of feedback effects we either don't understand or we don't even know about, driven by a host of stochastic inputs, of whose history we have only the most tenuous and inadequate knowledge, modelling of which is beyond us. Yet we can confidently predict its future behaviour in reponse to a change in one parameter". Yet it seems to me that the Met Office and Climate Science in general are saying just that.

Feb 8, 2013 at 1:12 PM | Registered CommenterMartin A

Martin,

which is my point that "natural variation" cannot be a null hypothesis either, because it's not a hypothesis.

Feb 8, 2013 at 1:18 PM | Unregistered CommenterTheBigYinJames

Jonathan, and TBY
Great comments. I was discussing the question of the null hypothesis with Mrs Shub (who is better at statistics than I am), since I noticed that the idea was being used by different people in a different sense than I am used to. The only 'null-alternate' idea that I am used to is the statistical one. I see that others use the term in different ways, and I have seen many discussions about p-values where different people are talking about different things (and everyone thinks they are right, and most are, but exclusively so which they are not).

For instance, TBY lists 6 different things as possible 'null' hypothesis. That is not the sense in which I understand the concept of null hypothesis at all. I think about it like how RKS did. I think both of you now see that there are these differences.

Secondly, if you read Jones' post, it is in excellent agreement with my characterization of lukewarmers as another brand of warmers (no insult intended here).

A quick note for now: there is ample evidence in the paleo record for abrupt climatic changes. And, there is no evidence to suggest that close to thirty-three years of research and work, deriving the physics of 'feedbacks', has brought us any closer to characterizing them any better. Clearly, something is wrong with the framework.

Feb 8, 2013 at 1:31 PM | Registered Commentershub

jeez, cross-posted again!

Feb 8, 2013 at 1:32 PM | Registered Commentershub

Lukewarmers are warmers, I think we all agree. Seems fairly self-evident fom the name :) The problem comes when that carries pejorative connotations. The implication being that if you're "with them" then you're "just as bad", a "watered down version" "lite ", "even worse than", etc etc. If I'm imagining this, then I apologise.

You're right, shub, we're using 'null' in different ways. When talking about an isolated parameter you're trying to work out the effects of, the null hypothesis is that it doesn't have an effect. The null hypothesis is a special case of a model hypothesis which says its effect is zero. The point of an experiment is to prove or disprove this. But it's still a hypothesis that needs very much to be proved or disproved like any of the rest. Some people are using the "null hypothesis" to mean "default explanation unless you prove otherwise", but this makes an assumption that the "null hypothesis" itself has been proved!

How the null hypothesis morphed into this hypothetical "I'm right until someone proves otherwise" argument is interesting but unrelated.

Feb 8, 2013 at 1:45 PM | Unregistered CommenterTheBigYinJames

We've had enough cross posts already!

Feb 8, 2013 at 2:02 PM | Unregistered CommenterTheBigYinJames

It has been interesting to see the different views on the meaning of a recently fabricated term - "lukewarmer". The term "null hypothesis" clearly has different interpretations in this context too. I would define the null hypothesis in this case with the statement that "anthropogenic carbon dioxide has not had, and will not have, any measurable impact on the Earth's climate".

In the absence of evidence to the contraty, the null hypothesis (as defined above) satisfies Occam's Razor, and is therefore most likely correct. We seem to be midway between LIA and MWP climatic conditions, which is exactly where the null hypothesis tells us we should be.

Feb 8, 2013 at 3:45 PM | Unregistered CommenterRoger Longstaff

Lukewarmers are indeed warmers: that's because (to misquote Margaret Thatcher) "the facts of physics are warming". But Lukewarmers are rarely warmists. Who you choose to insult is of course a private matter for everyone; personally I find that insulting people cuts down on drinking companions, so I try to restrict it to the truly deserving.

Personally I don't find the "null hypothesis" a useful concept in this discussion. The "default explanation" is a bit more useful, as it allows people to incorporate elementary physics into their statements, but if you really want to understand a Lukewarmer ask them for their climate sensitivity probability density function and how they get there.

Feb 8, 2013 at 3:47 PM | Registered CommenterJonathan Jones

Roger,

The trouble is, you can't really prove a null hypothesis, strictly, only disprove. It's there to be a counterpoint to the other hypothesis you come up with.

I like your null hypothesis on the CO2 question, because it neatly sidesteps the thorny issue of "can CO2 warm an atmosphere" and replaces it with the more simply answered question "has CO2 warmed our atmospehere?"

Feb 8, 2013 at 3:50 PM | Unregistered CommenterTheBigYinJames

TheBigYinJames,

Actually a better version of the question is "has CO2 detectably warmed our atmosphere".

Feb 8, 2013 at 3:58 PM | Registered CommenterJonathan Jones

Yes, quite right. The effect may be small, or it may have been negated by something else.

Feb 8, 2013 at 4:02 PM | Unregistered CommenterTheBigYinJames

TBYJ: I agree that it cannot be proven, and in this case can only be disproven over decadal, or even centenial timescales. Therefore, we are left with Occam's Razor.

JJ: Although we all beleve in "basic radiative physics", it should not allow people to people to incorporate elementary physics into their statements while ignoring other contributing mechanisms (conduction and convection) in a complex, non-equilibrium thermodynamic system. Futhermore, the mathematical application of radiative physics in this area has been incorrect (Holder's Inequality - see the thread on 255K and the Moon).

But I agree with you - I prefer drinking to having rows!

Feb 8, 2013 at 4:11 PM | Unregistered CommenterRoger Longstaff

"has CO2 detectably warmed our atmosphere".

Feb 8, 2013 at 3:58 PM Jonathan Jones

I'm struggling to follow the discussion.

You either have noise (= random climate fluctuation unaffected by human activity) or you have noise + signal (= random climate fluctuation + an additional warmth component due to human activity).

(I think this probably assumes linearity - doubtful ...?)

Is there the slightest hope of detecting the presence of signal if you have no understanding of the characteristics of the process generating the noise?

Feb 8, 2013 at 4:28 PM | Registered CommenterMartin A

Martin A, my "default assumption" is that we have CO2 induced warming plus a wide variety of noise processes including random climate fluctuations. I think it is reasonable to assume linearity over small variations.

Can you detect signal if you have no idea of the characteristics of the noise? No, you can't. If the noise could take any form then anything you see could be noise. This is particularly true of the sort of signal we are trying to detect, as a slow trend is particularly hard to detect in the presence of highly persistent noise.

So should we just give up? Well that depends on whether you think we really have no idea of the noise processes. If we really have no idea then our likelihood function is flat, and our posterior distribution is just our prior distribution. If we have some idea about the noise then we would get some influence of data on the pdf, albeit only weak.

Fans of unit roots and fractional Gaussian noise should note that this argument cuts both ways: it is extremely hard to determine the underlying sensitivity in such cases, which means that the recent mild warming provides little or no grounds for believing that the sensitivity is low.

Feb 8, 2013 at 5:47 PM | Registered CommenterJonathan Jones

JJ - Thank you for confirming that.

In which case, the key question seems to be "what do we know about the characteristics of the natural fluctuations of climate in the absence of human-induced changes?".

Obviously, we can make things up and pretend they are true, but what do we really know?

Feb 8, 2013 at 8:22 PM | Registered CommenterMartin A

Know? Not much.

And I still claim we don't know that the concept of climate sensitivity expressed as degrees per doubling over centennial timescales has any validity. If it is not there it's no use applying all your fancy statistical methods, what you will see is the sum of every unknown.

Feb 8, 2013 at 8:32 PM | Registered Commenterrhoda

just saying thank you to the people who have helped pull this post through the "RKS v the world" nihilistic stuff. And I would suggest to Shub that sometimes having someone say to an obnoxious warmist, "turn the volume down, please" can work.

Martin A - please do not think I was trying to annoy you yesterday - i was trying to draw out a real argument, which you did.

Feb 8, 2013 at 9:51 PM | Unregistered Commenterdiogenes

Yin, your conclusions seem odd:

"Natural fluctuation" - no or small rise in the next century
"Interglacial rise" - small rise in the next century
"Increase in GHG" - no or small rise in the next century
Combination of all/any of the above - small rise in the next century

What constrains "natural fluctuation" to a "small rise". Why not a big rise or a fall?
And if you agree with JJ that sensitivity might be 3K or perhaps more (you haven't argued the number), and since CO2 might go well beyond doubling (ok the effect is log, but even so...) there might conceivably (on your reckoning) be 4 or 5 degrees of warming. Is that small?
And then if you combine a possible positive natural fluctuation with that 4 or 5 degrees, is it still small?
It seems to me that you rapidly become a warmist unless you reject JJ's 3K (I seem to remember you claiming a limit of 1K many moons ago).

Feb 9, 2013 at 1:46 AM | Unregistered CommenterBitBucket

This connects to my first comment:

----------------------------------------------------------------------------------------------------------------------------------------------------

[1] Poptech at WUWT:

I have no hostility towards English majors like Mosher. I do have a problem with those who take what he says seriously without knowing his background; Mosher’s educational background consists of BA’s in English Literature and Philosophy. His “scientific” background involves bringing MP3 players to market for Creative Labs as a marketing director.

It is no coincidence that an English major would be the one to deduce Gleick’s involvement in the Heartland scandal since it involved recognition of nuance in written language. How that translates into caring what he has to say on scientific issues I have no idea.

Mosher frequently posts nonsensical “riddle” like comments here and rarely responds to criticism. When pressed he usually disengages and runs away. His comment tone reflects a superiority complex, believing himself to be the true “rational thinker”. He also frequently throws Anthony and other skeptics under the bus at other websites.

I also suspect he was the one behind the fabricated Muller as a recovering skeptic meme in the NYT op-ed. Which was easily crushed. His commenting is largely manipulative and or trolling. Manipulation of skeptics is the goal of those associated with BEST and people like Judith Curry. They want to siphon off the huge number of people who became skeptical after climategate to their brand of “luke warmerism” just with no change in policy proposals. Unfortunately too many people are naturally inclined to “moderate” type arguments but only so long as you can stereo-type those who disagree with you as “fringe”, peer-pressure does the rest.

[2] Poptech at WUWT:

Jeez, I am very familiar with computer programming resumes and Mosher does not even list a language he is proficient in. He has no formal training in anything computer science related. “Written code” can mean anything, including writing scripts or customizing pre-made files.

His resume specifically says, “Developed computer simulation models for aircraft systems and war games.” and “Created and coded data collection and analysis systems.” That can mean anything and looks like resume padding. Why did he not list the language he “coded” in? Programmers who actually write software are much more clear and detailed about what they actually did. People who give ideas about what should be in a simulation can claim to be part of the “development team” that does not mean they programmed anything. You can claim to “create and code data collection and analysis systems.” by writing a script to extract data from a DB. None of which is a scientific credential let alone the backbone of a computer programming resume.

Only non computer science academics would fall for it. Conveniently this makes up the entire BEST team.

He explicitly sells himself as a “Marketing Consultant”. Marketing and Sales are the two most frequently used words on his resume.

Those who think they are programmers is one of the things wrong with climate science,

http://www.nature.com/news/2010/101013/full/467775a.html

Yes there is a lot of things wrong with selling yourself as something you are not and I do not consider an English major to have any relevant scientific credentials no matter how much he pads his resume.

---------------------------------------------------------------------------------------------------------------------------------------------------

The whole thread at WUWT is a Mosher disaster. A great many people seem to have come to the same conclusion. I wonder why that is.

The Mosher bash thread: http://wattsupwiththat.com/2013/02/01/encouraging-admission-of-lower-climate-sensitivity-by-a-hockey-team-scientist/

Feb 9, 2013 at 3:14 AM | Registered Commentershub