Tuesday
Jun302015
by 
Bishop Hill
Bishop Hill Why science is not enough
Jun 30, 2015 Academia Climate: Models There's an excellent take-down of the "evidence-based policy" movement at SciDevNet. Author Erik Millstone seems to have a pretty firm grasp of things:
...the relevance of...models is more often assumed than it is demonstrated. In the case of climate change, some computer models of the impact of greenhouse gases on climate might usefully approximate to global realities.
Science advisers often ignore or conceal key uncertainties when offering judgements, perhaps catering to policymakers’ preference for reassuring oversimplifications
...some stakeholders might claim a uniquely authoritative understanding of an issue based on evidence
Reader Comments (23)
Unfortunately, climate models have shown no skill. Suggesting they might be "usefully approximate" is flat out wrong, unless their usefulness is for some agenda and not their predictive skill.
"Why science is not enough"
I am surprised by your heading, Bish: it is confusing and perhaps misleadng.. In the context of the rest of the blog the word "science" is never defined yet the term is thrown around like confetti.
Science is not easy to define, but simply saying one is being scientific certainly is not science nor is simply producing models on a computer.
Simply put, Science is using a number of methods in a systematic way: making careful, objective systematic observations, making a model or prediction based on these observations and then TESTING the model or prediction.
Only when these steps are followed to their completion can any conclusion be drawn with a high degree of validity.
A big problem with climatology is that the testing of the models requires passage of time.
Of course hard decisions sometimes have to be made without much evidence to go on but such decisions are not science.
The question then becomes "On what do you base your decision?"
I took the article as arguing for the following:
“Sometimes there is no evidence that we're all going to die horribly but that's no reason not to poop yourself and panic.”
In acknowledging that non-scientific factors should go into policy making, he is correct.
But in arguing that non-evidential factors should go into policy making, he is incorrect.
More GWPF luke warmist nonsense.
Jenny Murray asked Christine Rice about her early career as a scientist researching “global warming.”
Rice replied:
I was amazed really by the inadequacy of what we had, because we’re talking about climate change which is over tens of thousands of years as opposed to the twenty years of data that we had. So in a way we were putting out a lot of ideas and not really having concrete scientific research to support it, and I suppose at that point I did lose a little bit of my spark, thinking well I could propose an idea and I could probably draft a thesis that would support it and yet I wouldn’t really convince myself necessarily.
http://blogs.telegraph.co.uk/news/jamesdelingpole/100097342/christine-rice-your-new-favourite-mezzo-soprano/.
Models should never be considered until the models are 'proved'! Any serious privately employed programmer and their superiors expects thorough model run dissection after every run.
When the model runs are presented, those responsible for the models are available to explain why and wherefore.
In the CAGW world where models are given credibility without validity and model programmers neither back up results with detailed run information nor suffer for repeated failed model runs.
Private industry does not tolerate such incompetence.
Then, in the world of horrid non-validated unverified CAGW climate models, CAGW religious advocates insist of following maximum precautionary principle "just in case" actions.
CAGWrs believe;
Polar ice is melting, but neither ice cap is near ice free more likely to be ice free any time soon.
Somehow, without proven melting icecaps, the sea is catastrophically rising. Only sea level isn't
rising any faster nor is ice melting to raise the seas. Bangladesh is and will remain a swampy land.
Then there is the "passage of time", twenty years of bad models has sure proven how horrid the models are.
Climate models encapsulate all that the academics know about the climate.
The models fail to predict what actually happens in almost all regards.
From this the logical conclusion is that the academics don't know much about the climate.
But, the academics will not admit their models do not work.
From that we can conclude that they do not wish to admit how little they understand the climate.
Trying so protect the lives of children whose grandfathers are not yet even born, at the expense of the present generation, is akin to the Victorians having tried to foresee and solve all the problems of the 21st century. Just as we would have been furious if the Victorians had squandered their wealth in such futility, so our great-grandchildren will curse us for our stupidity in squandering their legacy.
But now that “climate change” is so firmly embedded in the public psyche, I don’t think “science” on its own will ever be enough to directly influence government policy.
As we know, politicians will say and do just about anything to get re-elected, so they’re on to a winner with “climate change.” It fits so many agendas. Most politicians are pretty smart, so even if they can see through all the nonsense spouted by climate scientists, they know this ideology commands considerable public support, so they are well-advised to support it if they want to keep their jobs.
Which is all a rather long-winded way of saying that, even if the science is proved to be dodgy, I wouldn’t expect much to happen until public opinion changes.
So my rather gloomy conclusion is that nothing much will happen until the public wakes up to the fact that they have been duped. I somehow have a feeling that Mother Nature may have a role to play here.
esmiff, I'm a bit confused.
Christine Rice's words seem very reasonable to me in this context. The telegraph link about "her early career as a scientist researching global warming" currently leads nowhere useful after a re-direct. What has this got to do with the GWPF?
The basic flaw in computer modeling isn't the computer, nor the model, it is the programmer. You can't write a program that is going to do "analysis" on anything without including your personal bias. The program starts based on the assumptions that the programmer considers valid. If they are not valid, it doesn't matter to the program and it will calculate it's output based on what it is told.
Thus, what is needed in this world right now are programmers that have no specific bias, and don't mind changing their programs to get them to reflect the truth, regardless of which way the changes need be made. If you hold to specific principles, such as red is purple, yellow is fuchsia, brown is chartreuse, and can't change those principles, then your program results aren't likely to have the correct output, but the output should be colorful nonetheless. Having said that, if you start your program based on carbon dioxide is going to turn Earth into Venus, then there is no way anything can change the output since that is "hard wired" into the program.
Are there any literate programmers that don't believe that carbon dioxide is the driver that are willing to take the opportunity to "code" a model that reflects what is truly known at this time, of the climate? Just the parts that are known, and then you can try to add "factors" that might also affect the outcome later to try to bring your model in line with reality. You can't "model" what you don't know, but you can attempt to simulate it better if you want to until the knowledge gets full enough to attempt to model.
michael hart
Sorry, I put a full stop at the end.
http://blogs.telegraph.co.uk/news/jamesdelingpole/100097342/christine-rice-your-new-favourite-mezzo-soprano/
Christine Rice was basically saying there is little basis in reality for climate science. Unlike the GWPF who want to argue about the fine tuning of computer models. You can't have a model if you don't have data. The models are useless for predicting the future. Arguing about climate sensitivity is akin to arguing about the fastest breed of unicorns . It's meaningless.
All models are wrong. Some models can be useful.
The progressives have appropriated computer modeling, which was once the sole domain of real scientists and engineers, and have corrupted it beyond all imagination. Real scientists and engineers use models to understand how things work, and how they might fail, in order to build them better and more robust.
Real engineers understand that there are LOTS of uncertainties in models, which must be accomodated by incorporating "margin" in a design. The amount of margin included is proportional to the risk, and inversely proportional to the cost. Everyone likes to have lots of margin, but no one wants to pay for any of it. Achieving a balance is something that comes from experience, sometimes very painful experience. No one wants building to fall down or airplanes to fall out of the sky, but no one wants to pay for an airplane that absolutely cannot crash (it would not fly), or a building that would never fall down (it would be one solid block of unobtanium)
Progressives have created the excreble "precautionary principle" to enable them to use models to generate scary stories that they then use to control the populace. The "precautionary principle" is a total bastardization and misuse of the concept of margin. The people who use these scary models do not talk about uncertainties, or about how they are just silly extrapolations of incomplete data sets which should not be extrapolated, because they are very hard(nigh on impossible) to model well(see, e.g. Navier-Stokes).
And the progressives use all sorts of social "science" to feed those models. Data from "experiments" that can never be reproduced, or that only applies to large populations and cannot be used to make any testable prediction about individuals, but which is used to justify regulatory decisions that will "save human lives". The public health field is rife with these, nowadays.
I used to supervise a group of engineers who ran models to inform decision-makers about nuclear power plant safety issues, and we ALWAYS informed the decisionmakers about the uncertainties, and how difficult it was to use those results to make decisions where the uncertainties were high. And they were very high whenever there was a human element involved in a scenario, which was most of the time. When the decisionmakers then make decisions about dangerous technologies based on how their politics makes them feel, or when they feel pressured by politicians to make the right decision, then things start to get really scary.
We used to compare our model runs to actual experiments and to plant transients to understand how to create better models, but I think it is fair to say that everyone was skeptical about everyone else's models. Deep down, they recognized that none of the models really represented "the truth". Escept, of course, for the people at the very top, who could claim that they could model 3-dimensional phenomena in a 1-dimensional model, because they were so great and skilled. All the rest of the modelers believed only the results of the experiments and actual plant transients because they actually occurred. Of course, the experimenters all had serious doubts about their own data - they knew what went into the design of the facility and instrumentation and they appreciated the uncertainties of the data. But they didn't talk about it much, unless there was a need for more funding and new projects...
rxc
There is nothing wrong with models. There is nothing wrong with climate models as a scientific tool, but when rascals like Mann and Schmidt telling us they can predict the future, we have entered the territory of the bare faced lie.
Freeman Dyson
I have studied the climate models and I know what they can do. The models solve the equations of fluid dynamics, and they do a very good job of describing the fluid motions of the atmosphere and the oceans. They do a very poor job of describing the clouds, the dust, the chemistry and the biology of fields and farms and forests. They do not begin to describe the real world that we live in.
http://www.edge.org/documents/archive/edge219.html#dysonf
I'm struggling with this bit (at the top)
What does it mean?
" A big problem with climatology is that the testing of the models requires
passage of time."
I don't really agree Morley. If you really wanted to test them you could verify each component of the output and see if they work. ie are the clouds, rain, ocean temps, ice, wind, etc. currently correct for every local region. You would then work out the importance of each bit and then you could probably drop the unimportant bits to simplify the model.
It is obvious they are completely wrong now for instance.
Rob, a question plus a comment.
Does not the verification of each variable in a model require time in two ways: the hard work in testing each model of each variable and observing whether the prediction of the behaviour of each variable mimics that observed at some future date?
There is a book written by an engineer "Structures: Or Why Things Don't Fall Down" Paperback – July 8, 2003
by J.e. Gordon (Author) that I thoroughly enjoyed. It describes the importance of models and also identifies many models that failed when inadequately tested.
Typical engineering use involves modelling of linear systems. Uncertainty can be tracked through the model. In the oil industry we use highly complex 3D models. We have understanding of the basic physics of fluid flow, which is largely linear, but not of the geology, which is very undersampled even for a field with lots of wells and seismic data. We need to use large stochastic simulations, even then we understand its only a model and has serious limitations. But our models are tested against reality constantly and we have to put our own money where are mouth is - not other peoples public subsidies.
So then we look at climate models. Modelling a non-linear, chaotic dual coupled system with no possibility or working at a fine enough scale. With huge swathes of unknown and critical physics missing. They do not work, climate models. The proof of this is the poor predictions and failure to fit past natural events eg warming to 1940.
Why is thearticle copyright marked "David Rose" ?
@Jack "Revealing assumptions will help science truly contribute to sustainability", What does it mean?
- He means when you tested a drug only on a mouse , then you say that you assumed 'if it doesn't kill amouse it won't kill a human'
sustainability is thrown around like confetti.
@Morley Sutter, says "science" is never defined yet the term is thrown around like confetti." - note how the story throws around the word sustainable similarly
* Surely You first test a model against knowns like the past, and then against unknown future.
Seems to me Jenni Murrays husband is a Greenie with hippy connections
MikeHaseler Jun 30, 2015 at 3:43 PM
+ 10!
Marginally off-topic - but what with the present warm weather I'm wondering how long it is before someone on the telly comes out with the dreaded phrase 'climate change'....
Which also makes me wonder how much of a field day the 'alarmists' would've had if they'd been around in 1976....
Having ,lived through and enjoyed it, was wondering that, too, Sherlock 1. I suspect tha if the climate scientist of the era were asked it would have been put down to 'Global Cooling,' all the rage at the time!
I have tried in the 'Discussion' section of this website (Jun 27, 2015 at 10:13 PM | Registered CommenterDung) to make the case that today;science is an unreliable source of advice relative to the world we live in. ^.^
The basis for that claim is that (relative to all the knowledge need to understand our world) today's scientists know bugger all. Scientific knowledge is incomplete and it continues to change with each new paper.
I totally agree with MikeHaseler Jun 30, 2015 at 3:43 PM and Jul 1, 2015 at 12:08 PM | Registered CommenterAlbert Stienstra.
" Does not the verification of each variable in a model require time in two ways:
the hard work in testing each model of each variable and observing whether
the prediction of the behaviour of each variable mimics that observed at some
future date?"
We can see that critical factors such as cloud cover don't work in the present so therefore just that makes them useless for what is basicallyly an energy balance question. As far as I'm aware the models don't conserve energy anyway.
The weather models which is all the GCMs really are useless for these long term forecasts.
And then there is physics, e.g. you don't need complicated models to figure out what will happen. That has been well known for over 100 years. They help with the details and figuring out the advantage of different policy changes