Have models received an unfair amount of criticism due to lack of understanding?

Question

I used to be incredibly skeptical of computer climate models. Mostly because I didn't know anything about them. I've since recently moved into working with complex, mechanistic models and I lose a bit more skepticism day by day.

While it may have been a duh moment, one of the things that greatly changed my mind was seeing the complex set of functions and differential equations that go into producing model output. I guess when i thought of a 'model' I just thought of an overly simplified computer program, completely devoid of calculations. Do you think that the adamant skepticism towards climate models is from a lack of understanding about how they work? Do you think that computer models need a P.R. boost and a name change, maybe something a bit more accurate and direct like "overwhelmingly large set of equations that would take an average person a life time to compute"

Do you think if the average person knew the work that went into designing these models that they would be so willing to question their output?

Answer 1

Absolutely.



I've found that the less people understand about a subject, the easier it is for them to reject its accuracy. This is only logical I suppose; when you don't understand how something works, it's easy to either become misinformed or make an incorrect assumption about it.



A good example is the causation of global warming. There's an immense amount of misunderstanding regarding what we know about causation. Most people think it's still hotly debated by climate scientists. Others make an assumption that because past climate changes were natural, the current must also be natural. Or they assume that we simply don't understand what causes climate changes.



But the more I've learned about how climate scientists attribute global warming, with radiative forcings and fundamental physics, the more confident I've become that the AGW theory is correct. It's the same thing with climate models - the less you understand about them, the easier it is to assume they're wrong. If you don't understand that they're based on fundamental physical formulae, that they're incredibly complex representations of the global climate based on what we've learned from thousands upon thousands of scientific studies, then it's easy to dismiss them as "garbage in, garbage out", or something similar.



The more a person knows about what goes into climate models, the less likely they are to dismiss their results. Certainly climate models will never be perfect, nor will any model, but they do give us a very good idea how the climate will change in response to a continually increasing anthropogenic forcing.



I'm not sure how much good a PR boost would do them though. Every time a name change is suggested for something AGW-related, deniers crap their pants.

Answer 2

In the end, nothing really matters other than how well the models predicted outcome compares with the actual event results, and it must have good results with multiple events with varying inputs before its accuracy can even begin to be measured.



Obviously, the models used by the IP CC are not yet very accurate since the actual data over the last 10-12 years has been way off. I assume with more time and more understanding of the numerous variable input algorithms they will improve it.



I personally believe we do not yet really understand the natural warming and cooling cycles well enough. When reviewing the historical trends from the ice core data, temperature rate of change varies wildly both up and down for periods of many years (10-40) long before man had any influence. I would think this known uncertainty would make it almost impossible to model with less than a couple of degrees plus or minus accuracy.

Answer 3

" It's only a model" displays the ignorance of whoever uses that as an argument, just like "It's only a theory" in a different context.



There is no difference in principle between a massive computer model, and solving a simple differential equation for a simple physical system on the back of an envelope.



And for many years now, model predictions have come with error bars. Fuel industry publicists pretend that this is the reason for inaction. Unfortunately, the very opposite is the case, because what this shows is that we cannot exclude some very unpleasant possible outcomes for anything like business as usual.



The public relations problem, I think, goes all the way back to a failure to teach people anything about the kind of activity that science really is, o rhow to think about probabilities. That and the notorious Dunning-Kruger effect (eg not knowing what the Dunning-Kruger effect is and not knowing that you really really do need to look it up and find out)

Answer 4

I don't think the average person would be able to come anywhere close to understanding just how complex they are. If they could, then they wouldn't be an average person, they would be well above average, at least in terms of mathematical abilities.



Since a good number of people go glossy eyed at anything above Algebra 2, the pages and pages of equations that go into a climate model will be completely meaningless. They might be impressed by the exotic symbols, but without a fundamental understanding of what they're actually seeing, it may as well be Chinese to them, and they'll just be as free as they were before to draw whatever simple conclusions they had in the first place.



The only solutions would be to make people more intelligent, or at least convince them to learn more math beyond what is relevant to their normal lives.

Answer 5

<<. Do you think that the adamant skepticism towards climate models is from a lack of understanding about how they work? >>



No, the skepticism has NOTHING to do with the actual science. None of the skeptics are skeptical because of scientific objections or questions.



Thinking that the minds of skeptics can be changed by showing them and explaining the science is like thinking you can cinvince the birthers by showing them Obama's birth certificate.



There might be a few exceptions, perhaps you are one, but very very few.

Answer 6

It's not about the understanding of math, but the processing power available. From what I understand it's not possible to stimulate two models running simultaneously with changing variables (basically real time) and have a accurate output. Anyway CESM is the latest greatest of the big 12.

Answer 7

Mechanical models are getting much better. However the science surrounding them is very refined and very well known. There is also a great deal of experimental and practical data available to further refine your model.



This is NOT true for climate models. There is very little experimental data available for climate modeling to use as ruler to gauge its accuracy. For example you can build a car or plane based on the model. You can then test the performance of that vehicle against the model, adjust the model to fit reality and then try again. This is called research and development.



For climate models, we have very little experimental data. Most of our reliable field data is less than 100 years old. for many of the moedel elements nearly all of our field data in less than 60 years old and in some cases less than 30.



Global climates shift too slowly and span far too much time for our reference frame to have statistical relevance when calibrating a model.



What you come down to is a very large set of assumptions. Not one single assumption is likely to 100% accurate and the composit accuracy of the model will be statistically lower than your worst estimate. (Errors compound, you do NOT average them)



Keeping it simple, we can NOT build a world based on our models and then observe how that world's climate responds. This eliminates the feedback loops enjoyed with mechanical models to refine thier accuracy, which in some cases are getting VERY good.



Climate models are by nature several decades perhaps more behind your mechanical models in accuracy. As we learn more and put more feedbacks into the models to aid them in more accurately predicing next months weather, they will become more accurate in prdicting weather patterns (climate) for several year and in time perhaps even decades out.



Any controls designer or technical model builder will tell you that without feedback loops to keep everything togehter and anchored to reality, systems will run wild.

Answer 8

Two quick comments:



1. Yes a model can include very complex equations. However, if the equations are not correct, or some input values are off, then the output of the model would be inaccurate. A major problem that skeptics claim with models is that there are too many assumptions resulting in different models having very different outputs. And then the scientists discard some of the more outrageous results and average the rest.



2. The climate models are based on weather models which are basically fluid dynamics engines projecting future flow. As far as I know, climate scientists took these models and altered them to assumingly try to mimic climate instead of weather.



I'm not sure why you working with models now would give you any greater confidence in climate models. If anything, it should give you less confidence.

Answer 9

Do you know how to calculate the expected divergence of a multi-parameter computer model?



In a mechanical model, I'm guessing the parameters are few in number and fairly quantifiable. Even under such circumstances, models diverge--usually rather quickly.



In a climate model, there are many many parameters, most are VERY LOOSELY quantified, others haven't even been thought of yet.



"Complex" and "computer model" are two words that don't really fit together well.



There are no complex computer models that track reality in a reliable way at all.

Answer 10

From that and a general ignorance of science. Models are ubiquitous in science. An equation is a model of some observed behavior. Climate reconstructions are models. Newton's Law of Gravity is a model, etc.



=====



PARTHA R --



>>go by theory more than ground reality. <<



Empirical data are the input (fuel) that run computer models. Without variable parameters, a computer model is a closed system.



====



Steve --



You've never even looked at one, have you?

Answer 11

In this world there are many things which gets influenced by the Politicians.



A Character Certificate is issued by someone whose character is in doubt.



Similarly, Models are designed by people not involved with a Project's ultimate objective. People go by theory more than ground reality. Therefore, things cannot be delivered properly because wrong people are engaged and they try to drive round objects in square holes. Naturally people not aware, start criticizing the Model without knowing the root cause of delay, disruption and failure.



I hope i have been able to get my message across.

Answer 12

I can build a perfectly sound computer model that shows that an over-pressured Cibicides Opima gas sand will have a Class 3 AVO anomaly.



If I drill a Class 3 AVO anomaly in an over pressured Cibicides Opima section... I'll drill a dry hole and I won't find any reservoir quality Cibicides Opima sands, much less one filled with gas pay.



The problem with modeling is the geophysical principle of non-uniqueness. The problem with models is the simple fact that there are no unique solutions. The "math" can be 100% correct; for every set of geophusical observations there are multiple model solutions that fit the data.



Models are great heuristic tools... They aren't very good diagnostic tools.

Answer 13

the more you understand the more question comes to your mind which need answers sooner or later you end up knowing how to control the weather aka weather control which really existed many years ago...

Answer 14

There are way to many assumptions for models to be reliable. There are way to many unknown factors. What is the point in pretending that models can accurately predict anything when we don't understand so many things that are vital to understanding climate, such as the details of cloud formation with increasing temperature, how the sun affects various aspects of the atmosphere including cloud formation, or even what effect CO2 has on the temperature. They don't understand the effects of aerosols, etc, etc. I could go on and on. Models are simply the only tool that alamists have and they therefore exaggerate their importance. This is obvious to those of us who don't have an emotional attachment to it.

Answer 15

Absolutely. Moreover, if the average person new how much real observations had already supported the early model's projections they would understand that global warming is real.



People who can understand climate science all understand that global warming is real. There is no dispute among climate scientists.