(about "models aren't science" and semantics)
I stand by that statement.
Archaeologists use shovels, but does that a shovel is a tool just of archaeology? Of course not. Models are just what they are. They are tools used by a variety of crafts, scientific and not scientific. They have their uses and their limitations.
This is what I mean. The models are tools to help in understanding things a potentially making future predictions off of them. The models themselves are not evidence or theory, and certainly don't collect data. Those are the three big things of all scientific disciplines.
Why? Because they require a bit more than just a passing "impression" of information? That they require more like a detailed understanding of science?
No, because scientists add or decrease certain things to make the model work without considering that they may have the relationships wrong, or that there could be the perverbial rock that's knocking that bowling ball the other way (i.e. unknown variables)
Why not? What's the difference between what I'm doing here vs what a hindcast model achieves?
I'm talking about regarding models being used for understanding potential mechanisms.
Please provide an example. (I'm curious if you have anything to back this up, I mean, remember: we are talking about a DOE and the results to wind up with associated t-tests on the coefficients and f-tests on the effects along with assessments of robustness of the model, so I'm interested if you can propose to me scenario with some details.
I'll lean back on the Ptolemy system which made predictions about the positions of planets that weren't surpassed until 19th century telescopes started separating themselves. The Ptolemy system relied on an Earth centered Universe with the planets and the sun orbiting in complex patterns to account for some of their wayward movements. This system lasted any hundreds of years and made perfect predictions, and predicted events all the way into the past. It even predicted eclipses!
This was a mathematical model that used past events. It fit past events, predicted the future with extraordinary precision, explained most events we saw in the night sky (with a couple of exceptions).
The problem is that, although this certainly works, it was not connected with reality in he least, and it was only finally debunked over a thousand years later.
This is the very danger overrelying on models. I don't expect it's that drastic with AGW models, but t shows that molding equations to fit past data gives you an understanding of a single mechanism. But, if that mechanism is not connected with reality, it leads you into a very,very dangerous direction.
Also, the theory of continental drift. This isn't necessarily a full model since I don't believe it was ever discerned exactly how this could work, but it was demonstrated that continents could float on heavier rock This, of course, replaced by Plate Tectonics, a theory that, so far, works tremendously and supersedes the previous demonstrations.
You are spending so much time trying to find details that you can disagree with.
I'm just telling you where I disagree. These are positions I've always held about models.
I will grant the scenario I painted there was physically more "simple", but for much of climate we do know many of the effects quite well! Something as simple as the "lapse rate" which gives rise to some heating at certain heights in the troposphere is predicated on the kind of stuff you learn in Chem I (enthalpy of fusion for water). Granted much of the movement and other materials is extremely complex and requires far more computing power than we used to have, but now days we have lots of computing power.
Models using tropospheric hotspots are problematic since weather balloon and satellite data all disagree on the degree of heating in that part of the atmosphere. Only one balloon and one re-assessment agree with heating. Balloon data that doesn't collaborate with ground and satellite data, and re-interpreting incorrect satellite data.
There is no proof of this yet, so any model that uses tropospheric data is making a leaping assumption.
You see, no matter what I presented to you as an example will fall into the same category for you. You'll agree with IT but not with this other stuff. There's some arbitrary line in the sand you have drawn and have to keep re-drawing numerous times based on the examples given to you.
The simple examples you gave are agreeable because we know all of these things to hold true from repetitive experiments.
With climate models, while some things are reliable like winds and currents, some stuff isn't, like volcanic feedbacks, albedo, the absolute effect of cloud cover, ice extent, and other things on one of the paper you gave me regarding feedbacks.
I don't claim we can ever get the models exactly right, but what I'm asking for are solid predictions. I pointed out that scientists aren't as confident as your article said they were (two posts ago).
2. Hindcasting is how we know how "robust" the model is (the Grida.no citation). You really dislike the hindcast example because you can't understand what the value of it is, but this is precisely the value. In order to claim that the models are a bust you have to overcome the fact that the models can and DO explain how climate works. If they didn't the FIRST thing they'd fail at is the hindcast data. (Are you having trouble with the term "hindcast"? Maybe that's the issue).
And, again, just because your model accurately fits the past doesn't mean it is a representation of reality. Ptolemy System.
"one mechanism"? I don't think you are following this at all. Please let me point you to a simple description of how climate models are developed from real scientists who do these models:
http://www.realclimate.org...
I apologize for my miswording of it.
I mean one set of mechanisms. Perhaps that is better. I'm merely pointing out that , again, the accuracy of the model to fit the past means nothing by tself. Again, Ptolemy.
(links)
Now do feel free to find all the "doubt words" in there and leverage that to assume all this science is bad, but it is far and away much better than the desire for total ignorance that the skeptical side seems to be going after.
Not bad, but not satisfactory for such confident conclusions to be drawn.
The Ptolemy System worked well too.
The failure of ptolomy's cosmology was that in order to make sense of how things worked increasing levels of complexity had to be added on an ad hoc basis. Ultimately the idea of the heliocentric system earth is part of simplified the system and proved superior. But note the key factors here: the ptolemaic system was undertaken at a time when science was hardly what it is today. To compare the kind of doubt in modern science to the wholesale ignorance of the pre-Copernican revolution is compare apples and oranges.
Again, AGW science isn't that bad. In fact, it's really not bad, but you do see the same sorts of things put into models to make them work. Add CO2, assume tropospheric temperature, and assume other variables. You have to tweak things to get them to work.
Yeah, the Ptolemy system ad to keep adding complexities, and AGW doesn't necessarily add complexity, but they still do change current variables.
Again, (sigh) that's why I put up the HINDCASTING citation. Does it kind of start to make sense yet? What am I missing here to get it through?
Hard to see where our misunderstanding is taking place.
Your model might work with for conditions in the past up until the present, but, a different model might be able to hindcast accurately in multiple ways with completely different feedback assumptions and variable values.
Hope that clears up our misunderstanding.
"All science is 'wrong.'" ~ drafterman