On Apr 1, 2:49*pm, "Ed Huntress" wrote:
wrote in message

...
On Apr 1, 8:11 am, "Ed Huntress" wrote:

wrote in message

....
On Mar 31, 8:54 pm, "Ed Huntress" wrote:

wrote in message

snip







I have a well developed BS meter and it is pegged.

he climate models are based on numeric solutions to partial
differential equations. These equations involve multiple energy
domains including chemical, thermal, chemical and fluidic. The system
is about the worst numerically conditioned possible with pressures
that range from 0 to 100Kpa, densities that range from 0 to 1000 Kg/
(m*m*m) with horrible geometry. If the world was a billiard ball the
surface features and most of the atmosphere reside within
manufacturing tolerances. The first thing I do when modeling a
dynamic system is determining the energy resides. For the atmosphere
system energy by far resides in the ocean where geology features have
a large influence on flow and temperature differences are even
smaller. That the physics is nonlinear, time varying and lightly
damped. I can't think of a worse problem.

I have some of the world’s best Computational Fluid Dynamics (CFD)
experts available to me and they have a rough time reliably modeling
turbulent flow reacting to moving structure on a wing. A recent
project with turbulent airflow over a simple 2 D wing had 10 million
nodes. The Godard climate model has 28 air/ water vertical layers.

================================================== ======

Well, that's good. So we'll put you down as one of those who doesn't
think
the models are useful. Most climatologists do.

Why don't you go straighten them out at NOAA?

--
Ed Huntress- Hide quoted text -

- Show quoted text -

Talking to NOAA and NASA would be like ****ing in the wind. When
someone publishes anything contrary to the GCC party line one is
branded as a kook or an oil company charlatan. It would cost me my
career. Talk about a religious fervor.

I don’t trust zealots of any persuasion and we have zealots making
policy. My state is making economic decisions based on model
predictions twenty years in the future.

It scares me that most climatologists think the models are useful. It
must be some kind of group think.

==============================================

It's possible. It's also possible that they know what they're talking
about.
And, based on the historical record of science and scientific
organizations,
I'll go with them over any individual, any political party, any ideologue,
any religious leader, and any writer of popular non-fiction books that
make
the best-sellers lists.

We all judge these things from our own experiences and prejudices. And we
judge others' judgments based on our assessments of them. I never got
involved in modeling, in engineering, biological science, economics, or
otherwise, but I've worked with experts in all of those fields. Often I've
had to interview them, prod and probe them, and analyze as best I could
what
they were saying, what their biases and blindnesses appear to be, and so
on.

Based on that, and for the record, I fully recognize what you're saying
about fluid dynamics. I've studied it -- or tried to -- and I regularly
use
it as an example to Larry and some others about how far over *most* of our
heads the actual science of climatology really is. I've seen the models
and
the math. Based on what I've seen through my limited exposure, I know that
there is no way am I qualified to judge the science itself. And, as a
writer
and researcher, I'm not inclined to accept what any third parties say
about
the science, either, because those people tend to have their own biases
and
blindnesses. My career discipline forces me to go to original sources, or
not to draw conclusions if I can get only secondary source information.
And
I don't like to write "he said, she said" interviews.

Also for the record, I believe you know about modeling. I also can see
you're doing it in an engineering environment. Having worked with
engineers
for most of my life, I have a good detector for the engineer's dismissal
of
other sciences, and I know some of the reasons why. You have to produce
concrete and quantitative certainties -- or results as close to certain as
science and technology allow -- and you know where the modeling
enterprise's
strengths and weaknesses are in that regard.

Likely you would tear your hair out over medical and bioscience models,
which I was exposed to over most of the past five years. As an editor, I
corrected the writing, the logic, and sometimes the statistics of medical
researchers for peer-reviewed professional articles. One thing you learn
when you have had one foot in the physical sciences for a few decades, and
switch to life sciences or social sciences, is that they're looking for a
different type of conclusion. Their worlds are subsets of vast, often
unknown sets of variables. When a mechanical engineer is exposed to fluid
dynamics at a high level, he gets a taste of that, and it frustrates the
hell out of many of them. Engineers don't like working with environments
that contain large variables that sometimes can't even be identified. They
also don't like to work with competing, unresolved theories. Neither one
fits their deductive-logic mindset. Fluid dynamics is full of partial and
competing theories -- submodels of the models they have to construct. But
you still have to come up with a quantitative result that fits within a
narrow range of certainties, one that is much narrower than the windows of
uncertainty that most other science-based intellectual endeavors have to
deal with. The mental attitude that flourishes in engineering is
antithetical to moderate correlation coefficients, modest p-values,
metaphorical abstractions, and close-call go/no-go decision making in
general.

So I will remain skeptical that you're characterizing the state of climate
modeling in a way that's useful for policy purposes. It may be an
*accurate*
way, from your perspective about what accuracy is, and how well defined it
must be to satisfy your engineering ethic. But it's dismissive in a way
that
suggests you'd be dismissive about life science or social science modeling
because they can't achieve the levels of certainty that you work with.

In a way, the decision making climatologists are facing is a lot like the
ones that doctors have to face when they're deciding on a course of
treatment in a life-threatening situation and the clinical research
indicates there is no alternative that shows more than a 60% probability
of
success. It's probably better than that in climatology, actually, but I
wouldn't venture what degree of certainty they're actually working with..
Based on what I said about science and scientific organizations at the top
of this rant, I'll accept that they're probably doing the best they can,
and
much better than anyone else can do, in a situation that can not accept a
"no conclusion" response.

We can note that the typical engineer's response to the data would be
"don't
build it." Unfortunately, it's already been built.

--
Ed Huntress
The FDA requires a medication be proven safe and effective before it
is approved for use. * * For the most part I agree with this high
standard because we are dealing with human life.

It's a sliding scale, not an absolute. If you're trying to get a fat pill
approved and it winds up driving one person in 5,000 - 10,000 to suicide
(rimonabant, marketed as Acomplia in Europe), the FDA won't approve it. If
it's a cancer drug that extends the life of late-stage terminal patients,
but it kills one person in 20, they'll probably approve it.

And that's based on a clinical study of a few dozen people with p = 0.1 or
so. Lousy accuracy. Lousy "proof."

That's much less precise than the stats on CO2's involvement in global
warming.

GCC is also dealing
with human life at a much larger scale. * If greenhouse gasses are
indeed causing significant climate change, then reducing them to
significant levels is going to kill a lot of people. * First order
analysis says that if one wants to reduce the output by 80% one must
decrease the input by 80%. *I can see wars over this because we will
have to tell the second and third world they have no hope of achieving
our life style.

I think your imagination is running away with you. The relationship in the
models is not linear. There is feedback all over the place, mostly positive
but some negative, at different thresholds. It looks like a lumpy curve.

We need to be FDA sure before we act or we will cause major harm
without helping.

Most of the world of science, and most of the world's people, seem to think
you have the Pascal's Wager part of this upside-down. The consequences of
serious global warming are more likely to lead to conflict, not to mention
that they're more likely to lead to physical hardship.
That's why governments and other institutions are worried about global
warming.

Dubos climate models won’t do that. * I would like
to see the topic pulled away from the political world into a frank
discussion of our options. * For example mitigation strategies like
not building on flood plains and moving food production might be more
effective then cutting energy consumption.

It might be. And if you follow the professional literature, you'll see that
those subjects have been discussed and analyzed endlessly.

I'm getting the feeling ...

read more »- Hide quoted text -

- Show quoted text -

It works with the FDA because it can evaluate the product with tests
against people. Get enough tests and one can generate meaningful
statistics. It is hard to get similar climate statistics because we
don’t have 1000 earths. If the FDA worked like the climatologists it
would simulate a human, evaluate the model against a human, and
generate statistics based on the model then teat the same human.
When I develop a dynamic simulation I never trust it until I evaluate
it against actual data. The desert is scattered with reasons why one
should verify every simulation. Bottom line: If a model hasn’t been
verified against real time varying data I don’t trust it.