“.. those subjects
who tested highest on measures like “cognitive reflection” and
scientific literacy were also most likely to display what he calls
“ideologically motivated cognition.”
You will not regret to
spend some time of this weekend in November reading and listening to
these two suggestions..
this article in
Nautilus, followed by listening to
this Econ Talk debate between
Kerry Emanuel and John Christy (about 1 hour, all of it worthwhile
listening), representing the main stream and sceptic view on climate
change, respectively. The debate is a good illustration of the
subliminal biases that burden our perceptions and opinions. All of
us are to some extent deceived by what we want to believe, and the
climate change debate is no exception. I found the conversation in Econ Talk interesting ons several accounts. The debate is always
conducted in a very civilized way, with no hints of dismissive
comments on the scientific qualifications of the discussants. The
debate is moderated by Russ Roberts, an economist, who tries to
understand the structure of climate models in terms of economic
models known to him- an interesting parallelism although not always
quite right. But what I would highlight of this conversation is the
difficulty of presenting a totally objective view on climate change.
Both Emanuel and Christy are premier league climatologist, and yet
one can recognize instances through the conversation when both try to
slightly bend arguments, when they present facts in a light more
favourable to their preconceptions, or when they ignore or do not
address those facts that may contradict them.
For example, in one
occasion Emanuel asserts that 'we know the equations that govern the
climate system'. Well, this is certainly not untrue, but it is not
completely true either. I layperson may
interpret that, indeed, there are no gaps in our knowledge of climate
change. Closer to the truth is, however, that we think we understand
the basic principles, each one taken individually, that govern
climate physics, but it is obvious that we do not understand well how
these principles operate in combination. We know very well that
sea-ice melts in warmer temperatures and that wind-stress drags
sea-ice away, but we do not understand why Antarctic sea-ice cover is
increasing. If we perfectly understood the dynamics of
climate, all climate models would produce the same results, which
obviously is not happening. Actually, although we 'know the
equations' most climate models cannot replicate the easiest climate
variable of all: the annually averaged global mean near-surface
temperature in the present climate, let alone the temperature trends.
Similarly, Christy
insists several times that climate models fail to replicate the
observed temperature trends over the last decades. While this is very
probably true, from this we cannot infer that greenhouse gases are not
having now and cannot have in the future a very important effect on
climate. It only means that the models are lousy and probably wrong.
In which direction they are wrong is not clear at all, yet.
All in all, I found
myself agreeing more often with Emanuel, but certainly Christy's
arguments cannot be dismissed out of hand. In the end, the
conversation shows that the issue of climate change has basically
ceased to be scientific - and thus the utility of the IPCC reports is
becoming really marginal. To believe that the climate change issue is
still a scientific one is one of the main mistakes of the mainstream
climate research community, and explains the misguided responses in
the blogosphere to the recent article by Victor and Kennel on the 2C limit. The 2 C
limit is not a scientific question. It is about setting a political
horizon to achieve reductions in emissions of greenhouse gases. As
such, the 2C 'political flag' is failing to effectively attract
enough supporters, as Victor and Kennel very rightly pointed out.
The Econ Talk
conversation illustrates it very nicely. The climate change debate
is all about perceptions: the perceptions of how solid our knowledge
is and the perception of how large the risk may be. These
perceptions are not going to change in the next decades. Some will continue to see a rabbit and others will keep seeing a duck, independently of what happens or fails to happen.
Eduardo,
ReplyDeleteChristy's claim according to which "climate models fail to replicate the observed temperature trends over the last decades" and your assertion "While this is very probably true, ... It only means that the models are lousy and probably wrong. In which direction they are wrong is not clear at all, yet." I would not agree to. While it may be so, an alternative explanation for the failure could be that the models are not subject to all relevant forcings - in other world, the moddels may be wrongly employed, while being themselves ok.
"It only means that the models are lousy and probably wrong. In which direction they are wrong is not clear at all, yet." I don't think I understand this comment. I expect Emanuel was bringing the GCMs as evidence. Christy is saying that they can't be used as evidence, they don't work well enough to be helpful. Is that wrong?
ReplyDelete@ 2 Mike,
ReplyDeletethis is the point I was trying to make. Christy argues that the models cannot be used as evidence, but on the way he suggests that greenhouse gas forcing will not be as disruptive, merging both arguments. I think both are different issues. Models may be very wrong, and nevertheless GHG may be very dangerous.
The bottom line is hat we have not understood the temperature evolution of the 20th century well enough yet . For some, like Emanuel, it is not that critical to recognize the risks of GHG gases. For others, like Christy, we have first to completely understand and then predict.
@ 1 Hans,
ReplyDeleteyes, I agree that this is still a possibility. But the deficiencies of climate models are not only restricted to the simulation of past trends. My comment was more general, also related to the problems they have in simulating the observed mean climate.
But assuming that the reason is a bad prescription of the forcing, the argument remains that models have not really been tested in a predictive sense.
@Eduardo
ReplyDelete"Actually, although we 'know the equations' most climate models cannot replicate the easiest climate variable of all: the annually averaged global mean near-surface temperature in the present climate, let alone the temperature trends."
Not sure if that is a good description of the problem. Models (very probably) could be made to score the globa Temp target exactly, however then probably many other things are worse than they are in the now disctributed version of theses models. The tuning range is very large and just one condition could probably be fulfilled easily.
For a more complete tuning philosophy see
http://www.iac.ethz.ch/people/knuttir/papers/knutti10cc.pdf
Also obviously after some work models do reproduce decadal trends (for the right reason is another question). The decadal trend problem is so far from the "do we know the equations problem" that I tend to see it as a science nearly apart. Progress is made by assimilation techniques or available ocean data. If you use or not a crap model to make the decadal prediction is rather a second order problem.
"To believe that the climate change issue is still a scientific one is one of the main mistakes of the mainstream climate research community,"
Like any political problem it has many aspects, scientific ones as well. A considerable reduction of the uncertainties of climate predictions (by new satellite data or whatever else) would have an impact on decision making. From "this is a scientific prolem" to "this is a problem that has nothing to do with science" there is quite some space.
'most climate models cannot replicate the easiest climate variable....'
ReplyDeleteGeorg wrote:
'Not sure if that is a good description of the problem. Models (very probably) could be made to score the global Temp target exactly, however then probably many other things are worse than they are in the now distributed version of theses models'
Well, I should have then written: Models do not (instead of models cannot). This seems to me a way not to face the substantial problem: models are to a large extent still inadequate for the accurate predictions that society or stakeholders would require. This harks backs to the title of the blogpost. It seems to me not totally honest to underline that models are based on physics (or that we know the equations) without simultaneously indicating that models are still far way of reproducing reality
'Also obviously after some work models do reproduce decadal trends (for the right reason is another question). The decadal trend problem is so far from the "do we know the equations problem" that I tend to see it as a science nearly apart. Progress is made by assimilation techniques or available ocean data. #
I do no agree with you on this point. The CMIP5 models, as an ensemble, still do not reproduce the observed trends over the last 30 years. This is not a problem of model initialization. Data assimilation in this context seems to me a fudge factor - although very useful for other purposes.
'Like any political problem it has many aspects, scientific ones as well.'
What can science still has to say that has not be said so far ? Do you really expect a considerable reduction in the uncertainties within the next, say, 10 years ?
@Eduardo
ReplyDeleteI haven’t seen your reply.
“Well, I should have then written: Models do not (instead of models cannot). This seems to me a way not to face the substantial problem:”
I didn’t express myself very clearly. The “substance” is not to change “cannot” to “do not”. I thought it’s important to mention that there is not one big target “global temperature” and the models miss it though they try very hard. The tuning process is a bit obscure cooking routine and it optimizes many things. There is not one target. The fact that they miss global temperature (which is arguably one of the less important things they miss) is due to their trial to get so many things right (meridional overturning, Hadley cell, OLR, mosoons, etc etc). I thought it is worth to mention that. Emanuel mentioned the optimistic “we know the equation” sentence when explaining to the economist guy what the difference is to typical economic models.
“I do no agree with you on this point. The CMIP5 models, as an ensemble, still do not reproduce the observed trends over the last 30 years. This is not a problem of model initialization. Data assimilation in this context seems to me a fudge factor - although very useful for other purposes.”
I do not understand what you mean. Besides of discussions on forcings and besides of good arguments that models underestimate certainly variability on many timescales observation were imported into a model so that it reproduces at least part of the last 30 years (ie the relatively small rise of the last 10 or 15 years) without being further nudged into the direction of the observations (studies of Virginie and Paco etc). How is that not an initial value problem? Does that not reply if you start at least this model one million times and vary initial conditions over the entire range that has been observed in the last 50 years you would reproduce the trends of the last 30 years? Might be the CMIP5 ensemble is only too small?
“What can science still has to say that has not be said so far ? Do you really expect a considerable reduction in the uncertainties within the next, say, 10 years ?”
10-20 years, Yes. Why not?
Compare climate sciences with economics. Obviously in economy there are more such things as “schools” (Keynesian, Austrian, etc) that will look at the same facts but allways come to different conclusions. Still, Economics and their studies are relevant for politics (even if politicians have clear preferences). I found e.g. the discussion on the famous eXcel error on how much public indebtedness is still sustainable. Even Schaeuble was justifying Euro politics directly with a freshly printed science paper. Not that somebody (Austerity vs Keynes) changed after discovering the error his/her opinion completely but still the science mattered.
http://www.newscientist.com/article/dn23448-how-to-stop-excel-errors-driving-austerity-economics.html#.VIGOhifA4qA
In the same way there are probably deep convictions among politicians if the energiewende is necessary, not so necessary, not necessary at all. They will unavoidably pick certain scientific results to foster their convictions. Meanwhile science goes on, makes progress and sometimes the arguments of some schools will disappear (as Marxism basically disappearad in economics [besides of Spainhttp://www.elmundo.es/espana/2014/11/27/5477550722601db8758b4571.html ]) since they are no longer in agreement with any evidence. So for quite some time there will be a low climate sensitivity and a high sensitivity school (as they were proponents of public financial spending or austerity) and at some point in the future one of these school will obviously disappear (if in 10 or 20 years, I don’t know, but at some point it will happen). Until then the public discussion will find in science enough material to do the Talk (of the society, politics, moral etc). I don’t think that’s a bad thing. Science is part of how opinions are formed.
Georg,
ReplyDeleteI have the impression we are getting our wires crossed.
I am not making the 'tuning' argument. I am just stating that models as such cannot reproduce, as an ensemble, the observed global mean temperature, regardless of the tuning. I am specifically referring to the
this paper , where models are also at the fringes of observations - on of the reason why model sensitivity and observational based estimations of sensitivity do not agree. On regional and longer timescales, for instance if you compute the observed 20th century temperature trends at each grid cell and compare it with the simulated trends , plotting the differences of trends, you would need a range in that plot of -+ 2K per century. Are centennial regional trend differences due to internal variability ? Maybe , but then it is indeed impossible to estimate any costs of climate change. If those trends are indeed due to external forcing, then models fail, so far, as a useful predictive tool for policy making.
This is my whole point. I am not telling that anthropogenic climate change is not happening/ will happen. My point is that to insist that 'we know' the equations, or to insist that we cannot predict the system is by far not the whole story. Both commentators, Christy and Emanuel, are just telling their side of their story. The question for the social scientist is why?
'Science is part of how opinions are formed'
Here I must disagree with you. I do not think science plays any role whatsoever in the formation of opinions or policy in 99% of the time. Science is mostly used as a support for preconception. There are many examples of policy decisions not based on silence, ranging from bans on smoking and security belts - which were first prescribed without previous scientific statistical studies- to homepathic treatment (paid by Krankenkasssen ) to the phase-out of nuclear plants. Perhaps that 1% of the cases are the only ones in which human beings make the right decisions. Your point with economics as as science actually supports me in this opinion quite clearly. The skill of economic short-term predictions , say one or two years, is basically zero. There are glaring examples of all models failing to predict even the sign of changes in interest rates, exchange rates, etc. one year ahead .
'If policy makers would just understand the science....'
' If scientist would just understand how policy decision are made...'
“Both commentators, Christy and Emanuel, are just telling their side of their story. The question for the social scientist is why?”
ReplyDeleteBecause it is a discussion, Eduardo. You try to make a point and “win” the discussion without lying if possible. Emanuel underlines the physical basis of the models and Cristy the uncertainties of predictions. I have no particular problem with that and I don’t think a sociologist is needed to understand that. Still I think when explaining climate modeling to an economist it is fair to mention that there is a physical basis of the equations in a climate model in contrast to economic models. It is hard to explain a scientific tool/approach and to start with the uncertainties.
The tuning argument came from me and I wanted only to add something to your point that “the models don’t get the temperature right”. You mentioned global mean temperature trend. If GCM models were developed to get just this one number right it would be really an enormous waste of time, money and talents. Temperature is just one of the many things the models try to get right. That is no contradiction to what you said just some additional remark.
Could you explicitly add the link to your paper. For me at least it doesn’t work. So without haven’t seen the paper I agree with you that based on existing regional trend projections any concrete policy making for a specific region seems very difficult. Might be that is one reason why everyone is speaking about mitigation and not so much about adaptation. Adapting to what specifically?
“I do not think science plays any role whatsoever in the formation of opinions or policy in 99% of the time. Science is mostly used as a support for preconception.”
But the preconceptions about what to do with climate change are soaked into science in a similar sense as e.g. the decision of pope Urban II to start the crusades was settled in religion. Science is the (or at least: is one) ideological background when it comes to discussions and negotiation on mitigation. But obviously there is not exactly one precise way to less CO2 emissions as there it is not exactly written in the bible that the christians had to conquer Jerusalem. Still the latter decision had to do with religious believes.
“Preconceptions” are not bad as such. It allows to the people to order new information into a general picture. You put the weight on the apparent immobility of the preconceptions. I would rather look at the slow evolution. As long as the ideological background is affected by science views and opinions will change when the science changes.