[Osx-nutters] Denial

[Stefano Mori]

On 2007-Aug-15, at 20:38, Chris Gehlker wrote:

>> But the thing is, having a complex model doesn't
>> necessarily mean that your predictions are more trustworthy.
>
> OK, but I still think you are missing something fundamental. The GW
> proponents *did not start from a complex model*. They started from a
> dead simple physical model that can be tested in any  middle school
> science lab. That  model is verified every time someone sees a
> hothouse tomato. The greenhouse effect in the lab is simple, easily
> testable, and not disputed by anyone.

I've felt the greenhouse effect inside a greenhouse.


> The problem arose when they tried to apply this simple model to the
> planet and discovered that the observed temperature rise of the globe
> was less than what would  have been predicted by the lab results.
>
> Now this isn't very shocking. Most systems out in the world don't
> behave exactly as theoretical physics predicts they will and in
> accordance with lab experiments.

I've never felt an ice age inside a greenhouse.

Anyway, that's just by the way.

> But is does seem to me that it  is
> qualitatively different to start with a model that you can verify in
> the lab and then work work toward something that's a more accurate
> predictor on a global scale than it is  to work on something like
> smoking/cancer where all you have initially is a correlation and you
> are trying to figure out the mechanism as you go along.

No matter what the subject, people seem to easily slip from having a  
model about mechanisms based on experience, to having opinionated  
predictions about the future.

Just because you have a model, just because you've studied  
mechanisms, just because you've gathered data, doesn't in itself mean  
you can predict the future reliably.

Let's say I'm in ancient Greece and I have studiously measured the  
depth and span of all stone lintels of varying sizes. I model the  
relationship and theorise a mechanism regarding the strength of  
stone. Then one day I'm asked to calculate how deep a lintel has to  
be for a wider than normal opening. I use my model to predict the  
required depth. I say the prediction can be trusted because it is  
based on thousands of measurements and an understanding of the  
mechanisms. They build it but the stone is now so heavy that it  
shears under it's own weight and collapses.

Even if your climate model can roughly follow current observations,  
when we model doubling of co2 we're operating outside of our sampled  
observations. So how the heck are you going to demonstrate that the  
model should be trusted?

Stefano