B
Bill Sloman
Guest
On Sep 27, 5:26 am, dagmargoodb...@yahoo.com wrote:
phyiscal system - they are all more or less useful simplications.
undestand what they are doing. The art lies in getting a useful
approximation to reality.
seriously, or waste time concoting such a model, but it does make a
comical straw man.
As I said, you are effectively claiming that a large chunk of academic
research can be written off on the basis that you think that they are
making the kind of mistake that you presumably made when you were a
wet-behind-the-ears newby in modelling business.
It's definitely comical, and entirely pathetic.
--
Bill Sloman, Nijmegen
You don't understand the details. No model accurately represents aOn Sep 26, 6:21 pm, Bill Sloman <bill.slo...@ieee.org> wrote:
On Sep 26, 9:59 pm, dagmargoodb...@yahoo.com wrote:
On Sep 26, 12:53 pm,BillSloman<bill.slo...@ieee.org> wrote:
On Sep 26, 6:06 pm, dagmargoodb...@yahoo.com wrote:
On Sep 26, 9:31 am,BillSloman<bill.slo...@ieee.org> wrote:
On Sep 26, 1:26 pm, dagmargoodb...@yahoo.com wrote:
On Sep 26, 4:20 am,BillSloman<bill.slo...@ieee.org> wrote:
On Sep 26, 6:22 am, dagmargoodb...@yahoo.com wrote:
Even a small error overshadows the entire alleged AGW signal.(**)
But most of the models predict roughly the right amount of cloud, and
fit the the sort of AGW we've seen so far.
Your own linked expert said the opposite.
Some of the range of models tested didn't predict the right amount of
cloud, most did.
o "Right amount"?
"...one of my
runs ended up with no clouds, other
people had all the water precipitate
as ice at the poles, etc.)."
Yep, they've got it nailed all right.
o Models were gauged by hindsight, not successful predictions.
o "...most did"? Where'd you find that?
In the bit of the quote that you first posted, then snipped when you
reposted it above.
Then you mistook what he said.
He said they ran a spread of models with various parameters, then
selected the model runs that correlated with PAST data. Producing
correlation with past events was the parameter-selection criterion.
That's adaptive curve-fitting.
In fact adaptive curve fitting involves adjusting the value of a
particular set of parameters to fit a particular set of data (which
can be quite large).
The spread of models with various parameters are a variety of
different models, rather than a single model being adjusted to fit
historical data.
You've failed to understand what's going on, and on the basis of your
imperfect understanding have written off a large chunk of academic
research on the basis that it's something simpler - and totally
inadequate - which you do think you understand. It's distinctly
comical.
You're lost in the details. If the model doesn't accurately reflect
the physical system, tweaking a bunch of unrelated coefficients
doesn't rehabilitate it.
phyiscal system - they are all more or less useful simplications.
That is the sort of mistake a beginner can make, if they don'tYou can artificially make it reproduce
arbitrary historical data, and pretend you've modeled reality, while
conferring zero actual power of prediction.
undestand what they are doing. The art lies in getting a useful
approximation to reality.
I imagine it could. Nobody in their right mind would take itThe same methodology, applied, could extract the pertinent parameters
w.r.t. to fires, fire intensity, and historical data regarding red
trucks. A model would quickly emerge showing strong correlation, with
coefficients capable of roughly predicting the size of the fire based
on the number and size of the trucks, plus other data.
seriously, or waste time concoting such a model, but it does make a
comical straw man.
Perfectly correct, and perfectly irrelevant.But, model or no, it's wrong--red trucks don't cause fires.
As I said, you are effectively claiming that a large chunk of academic
research can be written off on the basis that you think that they are
making the kind of mistake that you presumably made when you were a
wet-behind-the-ears newby in modelling business.
It's definitely comical, and entirely pathetic.
--
Bill Sloman, Nijmegen