On Fri, 13 Nov 2015 14:43:47 -0800 (PST), "
wrote:

On Friday, November 13, 2015 at 4:36:06 PM UTC-5, Ed Huntress wrote:

In my opinion this subject is still a matter of opinion, not fact.

That's because you don't know the facts, and you have no way to
evaluate them.

Until someone comes up with a model that agrees with the actual data, it is still an opinion.

No. You're misstating the nature of statistical models.

Even when someone has a model that agrees with the actual data, it is still a theory. Einstein's Theory of Relativity is a theory, even though many experiments have be made that show the experiment agrees with the Theory.

Dan

Theories are not opinions. Theories are based on facts.

--
Ed Huntress

Sorry but I still disagree. If everything is facts as you believe, there would not still be scientists mulling over the data. When things are well known and understood , there are not people studying in the field. Ohms law is an example. It is well known and there are no studies being conducted to verify the results. No government grants to study Ohms law. Lots of government grants to study climate warming.

Theories are based on facts and conjectures. Take the Big Bang Theory. Part observation , part conjecture.

Dan

"Conjectures," or assumptions, are part of stochastic modeling and are
NOT opinions. In a good scientific model, the assumptions are
themselves based on scientific facts. The reason they're assumptions
is that their specific, individual truth values may not be known. The
big issue then is whether these assumptions are determinant. The next
step will tell you if they are.

So when you have that situation, you test your model with different
values for the assumptions. My son runs stochastic models that may go
through 100 iterations like that. In climate science, it's probably
1,000 or many more.

The usefulness of your model in that case depends on demonstrating
that the alternate assumptions don't steer the model off in a
different direction. Hurricane-path models are an example. You may not
know if it will land in South Carolina or Delaware, but you know it's
going northwest. Rarely, it goes off in some other direction. But
you're limited in that case because you only have a limited time to
run your models, and the inputs keep changing. In long-term climate
modeling, you have the historical data and it takes years, typically,
for the inputs to change significantly.

This kind of modeling is not "opinion." And when you run a dozen
models with different assumptions and they all track the same path,
but with slightly different values, you know that the unknowns are not
determinant, but only constants (in the mathematical sense) that vary
the specific output values.

That's where we are with climate science. Look at that bundle of
snakes in the graph reprinted in the article we were talking about.
They predicted the directions, both up and down, that were followed by
the post facto data measurements. Those are good models. You will
NEVER get perfect predictions from a stochastic model, if for no other
reason than that it's statistical, and is based on sampling, not on a
universe of measured examples.

Contrast that with a mathematical model that predicts the fatigue life
of an aluminum airplane wing. You may never have perfect data in that
case, and you may not know the moment a break will occur, but you can
predict it very closely because you don't have a huge number of
variables, and their interactions are well known and consistent. You
can reduce them to mathematical formulas that can be tested and that
don't vary except with physical anomalies, like an internal stress
that can't be controlled. Not so with climate.

People who live with stochastic models recognize a good one. Climate
scientists live with them. And almost all of them recognize that their
models are all but incontrovertible. They can't tell you how many
degees it will warm in, say, 2050, but they have a high certainty that
they know the range. And it will be hell for some people on earth.

As for the other 3% of scientists who disagree: I'll take the 97:3
odds.

--
Ed Huntress