Sorry guys, but its kinda long-winded.
Emergent phenomena are important for 2 big reasons. First is that they're a check on whether your the interactions in the model are realistic. By definition, they're a consequence of the interactions defined in the model. You would be extremely unlikely to get the complex interactions wrong and still end up with numerous realistic emergent phenomena. And keep in mind, temperature trends are another emergent process.
Second is that they constrain tuning of the models. The output of the models is tuned to better agree with the data. However, as you change the parameters to change the shape of the curve, it also affects the emergent phenomena due to the coupling. While you might get the model output to fit observed temperature trends better by changing a parameter like the sensitivity to CO2, you also change the output elsewhere in other process, possibly making the fit worse there. If I fit the average temperature curve better, but no longer get realistic precipitation patterns I probably haven't actually improved the skill of the model (though statistical tests will determine that).
I guess we will never agree on this. It just doesn't make any sense to me that you are using accurate modeling of some aspects of emergent phenomenon as a justification for how these climate models must be able to accurately reproduce ANY phenomenon relating to climate. As a point of agreement and from a modeling perspective, yes, the abilities of these models are impressive, but there are so many systems, pathways, and interactions that aren't included in these models. We are, literally, learning everyday about new things that have been ignored in these models and their significance.
What if it is possible to recreate much of the emergent phenomenon if only submodel A,B, & C are included, but to accurately model the systems that come into play in the case of anthropogenic warming you need submodel D which has, as yet, not been included in the greater climate model? My point is that there exist systems that may not be needed to recreate much of the emergent phenonmenon, but they may play an integral role in the case of anthropogenic warming.
We seem to be talking in circles. The models naturally produce hysteresis. They're validated against past conditions which include both path "A" and path "B." The same models work for both paths. In fact, a decade before proxies showed the existence of path "A" with CO2 lagging temperature, it was predicted by models based on path "B."
I'm not sure exactly what you are referring to as path A and path B now. I was using those pathways as hypotheticals, not anything specifically. I assume you mean path B to mean anthropogenic warming. In what way are the current climate models based around path B (anthropogenic) when they are validated against and tweaked to fit prehistoric climate proxies?
Whether the significance is overstated is beside the point. The point is that there is no less requirement for models to be based in reality than lab experiments. There is no such thing as a model that is not based in reality. That would be a program, but not a model. Models are simplified abstractions of reality, but so are lab experiments.
Again, I disagree. There are a ton of models produced that have little to no basis in reality. A hot topic in physics right now is String theory. It should actually be called "String Model" in my opinion, because, as yet, it has no basis in reality and no testable predictions/hypotheses. Physicists are interested in it because it is mathematically "elegant", not because it has revealed something new about nature.
It's done all the time.
It's not a climate example, but one that immediately comes to mind was used in the case for making turtle excluder devices mandatory equipment. The trawlers argued that they were mainly removing old individuals and reducing density dependent mortality on younger turtles (which is sound in theory). In their minds, the reduction in turtle numbers was due to the reduced survival of juveniles due to beach development. Well, you can't stop trawling or beach development to run controlled tests, there's no historical data to use in a BACI tests, there's no practical way to isolate an undisturbed population of sea turtles, and you can't raise a population of turtles in the lab. Models are the only practical way to test whether trawling has a significant effect on turtle populations. Population models based on the real world data about turtle life-histories indicated that the population was fairly insensitive to changes in juvenile survival. However, even small changes in the survival of adults made a big difference in the population trajectory. There was no initial attempt to project actual turtle population numbers in the future.
That turtle model did NOT test the hypothesis. It was useful and insightful, but it did not test the hypothesis. Did the model somehow force nature to obey it's results? The only thing that could test the hypothesis is to actually monitor trends in the turtle population. I don't want to be rude, but you ought to review the tenets of the Scientific Method.
Some form of observation or experiment is an ABSOLUTE necessity in testing a hypothesis. Einstein predicted that star light passing very near the sun should be observed to be bent by the gravitaional field of the sun during a solar eclipse. That prediction didn't support the theory until it was actually tested by observation. The theory/model cannot simply support itself with it's own predictions.
It seems like you're trying to move the goalposts here. The original contention was that models are not scientific. We both agree that they can be used to make projections/predictions about a system based on observations. That's the first step in science. They can be and are also used to test hypotheses for which there is no practical way to test in the real world. You still seem to be hung up arguing about them having no basis in physical reality though, which is not the case. A program not based on reality isn't a model, and there is no way to validate it. I just grabbed a modeling textbook off of my shelf and the second sentence of the book is (emphasis added)- "Models are, by definition, simplified representations of reality." Yes, you can put unrealistic inputs into the model, but you can do the same with lab experiments (e.g. Miller-Urey).
Not moving goalposts. Models are not scientific in that they have no ability to test a hypothesis.
I think you might be misinterpreting what I mean when I say that they aren't required to have a "basis" in physical reality. Perhaps, I worded it poorly. I guess I should have said that models do not have to actually be faithful to reality. In my previous example with String theory, we have a model that requires 11 dimensions for which we have no evidence. Are there really 11 dimensions? We don't know. It is an intersting prediction, but it is currently untestable and may actually never be able to be tested due to the very laws of physics that it seeks to explain.
In medicine/toxicology, there are pharmacokinetic models that treat the body as a series of distinct compartments and defined sets of rules for how the compartments interact. Should we start performing surgeries to look for shoe box-like compartments inside of people? Of course not, they are just mathematical abstractions. Do the rules set forth on how the compartment interact actually represent all the complexity of the real human body? Not by a long shot. These models are useful, but like all models, they have a finite range of usefulness. They are good at toying with to providing insights into the utility of drug, they wouldn't be useful as the sole means to treat a patient walking into a doctor's office.
Lets say that I had a pharmacokinetic model for Penicillin. I am a doctor and I used a pharmacokinetic model to try to determine the effectiveness of using a particular dose regiment on a patient that just came into my office with bronchitis. I calculate the effectiveness using the model and it seems that the normal dose regiment would be effective. So, I go ahead an administer the Penicillin. After recieving the medicine, the patient goes into anaphylactic shock. What went wrong here? My model said the patient should be improving, not getting worse. The problem is that the model didn't include processes or inputs that address the possibility of allergic reactions.
I see climate modeling much like my hypothetical doctor's office visit. We are relying on models that may or may not include all relevant processes, to determine a "treatment" for the global warming "illness".
Again, with the Miller-Urey experiments, the inputs were not unrealistic. Nothing about the experiment was unrealistic. The experiments only showed that organic molecules could be created under a specific set of conditions. That is not to say that the interpretation of these experiments or the viewpoints imposed upon them might be unrealistic. The experiment itself can, in no way, be considered unrealistic. The experiments happened in reality and therefore would have to be realistic.
Or that it's a value that is difficult to measure directly. I can tell you all about what a carrying capacity is and what it's determined by, and even though, it's a real value, it's a product of multiple factors, so I can't just go out with carrying-capacityometer and get a value for it. I can try to identify and measure every environmental and demographic process that determines what it is, or I can use logic and relationships to constrain it. It's easier and usually more accurate to infer the carrying capacity from measurements of other parameters.
I would say that the "carrying capacity", as you have described it is, in fact, not "real". It is a mathematical abstraction and hence cannot be measured. If you can calculate the carrying capacity without making any assumptions about the processes that contribute to the carrying capacity or assumptions any of the values of the factors that define the carrying capacity, then you probably have a decent understanding of what the carrying capacity is and what it means (again, as a mathematical abstraction).
On the other hand, if you were forced to make assumptions about the value of carrying capacity, which processes affect the carrying capacity, or the values of the factors that comprise the carrying capacity, then you probably have an inadequate understanding of the carrying capacity because you do not have the data to provide values for the factors/processes that define the carrying capacity as a mathematical abstraction.
All models in every field are used with the complete understanding that they are all wrong! There is a reason you learn the models in school though. That's because they are USEFUL, even if they're wrong. Newtonian physics is wrong too, but unless you're traveling close to the speed of light, it's useful for most applications.
Agreed, but with the caveat that the models are only useful under a certain range of conditions. Knowing what the range of conditions for which the models are useful is quite possibly the most important part of modeling.
Read the actual climatology literature. You're complaint isn't borne out. They are quite clear about where more data is needed, where the models are unreliable, and how far out the projections can be trusted. There is an entire section in the IPCC reports dedicated to that discussion.
I have been reading as much of the climatology literature as I can find time for recently. This seems to just be a difference of opinion. I personally feel that documentation of the model's shortcomings is inadequate or underemphasized. If I remember correctly, this is in congruence with the comments of many of the reviewers of the IPCC documents. Part of the issue is that much of the climatology literature is published in scientific journals with the unspoken understanding that other climatologists who are already familiar with much of the models' shortcomings will be reading the journals. Hence, many of the articles do not go into the gory details of the shortcomings and the range of usefulness.
Scott
