From: Johnicholas Hines (johnicholas.hines@gmail.com)
Date: Mon Feb 23 2009 - 13:52:10 MST
On Mon, Feb 23, 2009 at 3:40 PM, John K Clark <johnkclark@fastmail.fm> wrote:
> On Fri, Feb 20, 2009 at 10:49 PM, Johnicholas Hines
> <johnicholas.hines@gmail.com> wrote:
>> Suppose a weather simulation can produce images
>> that can fool people who have seen weather.
>> Should you base your strategy against global warming
>> on that weather simulation? I think the answer is no.
>
> I think the answer is yes. If computer weather simulations weren't any
> good at predicting what you will see outside your window tomorrow then
> there would be no point in making them. But perhaps I misunderstand you
> and you were trying to say that a simulated heat wave will not heat up
> you computer more than a simulated blizzard; if that was what you were
> getting at then I agree.
Maybe concrete examples are in order?
Simulation A "FINITE ELEMENT CLOUDS"
Suppose this simulation predicts the temperature that the temperature
sensor will read, and the wind speed that the wind-speed sensor will
read, both accurately. However, people watching the simulation are not
fooled - they can tell the clouds are blurrier than real clouds.
Simulation B "ELIZA CLOUDS"
Suppose this simulation is very pretty, and nonexperts cannot
distinguish between images of the simulation, and images of real sky.
However, it doesn't actually have any "temperature" variable inside of
it, driving the phenomenon, just a big database of how clouds look and
seem to change.
My claim is I think we should consider the expert opinion that
"there's no weather underneath" as relevant in deciding whether to
base your global warming strategy on the simulation. The
"indistinguishable in outward behavior to the non-expert eye"
criterion isn't sufficient.
Johnicholas
P.S. Nesov also made a much better argument, more directly regarding
uploads which were changed by the process.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:04 MDT