In article ,
John Larkin wrote:
[...]
The chaos here is restrained by a larger, slower, but more powerful
overall negative feedback system.

Both these are examples of systems that robustly chaotic and yet the
average results are easy to predict.

But now consider whether your arm will be "up" or "down" at exactly
noon of January 6, 2008. If I had washed my car, the result would
likely be different.

Whether I raise my hand or not is not an issue as large as an ice age.
It is like the details of the switching of the hit and miss circuit. It
may not matter in the larger picture.

There are many implications to washing my car.

The biggest one is that it always rains after you do. Washing the car
may (only may) result in other larger things happening. There is a
greater chance that the effect will die away.

[...]
The Earth oscillates, very noisily and aperiodically, between ice ages
and jungle ages.

"oscillate" is the worng word if this is chaos. With chaos, you can't be
sure that the system will repeat or recover from an extreme. Chances are
the system has a restoring action along the lines of:

Initially, the snow is white. Gradually after the water has been taken
out of the atmosphere, it darkens due to dust etc. When it gets darker it
starts to absorb a bit more sun light.

Whether we'll be hot or cold 200,000 years now could
be changed by the tiniest extra input to the system. Certainly a major
change in solar output would change climate, but the noisy swings
about the mean, the chaos component, are large and exquisitely
sensitive to input.

I agree, it could be changed by a small input. The point I'm amking is
that a larger input is more likely to make a change. Things like a change
in the solar constant are going to win over you washing your car almost
every time.

Now consider a short time frame. Our weather forcasts work fairly well
out to the next day or so. This means that local changes can be predicted
to some accuracy over that period. You can predict small things in the
short term because the things that effect the short term things change
slowly enough.

Consider a slightly longer time frame. We are right now seeing a
temperature increase. Lets assume that this increase is because there is
more CO2 than normal (ie: assume the solar constant is) We can predict
that this warming will continue for some time because the CO2 can't be
increased or decreased quickly.

Now imagine that we have some method by which we can apply feedback to the
system. If we apply a positive feedback that increases the CO2 when the
temperature is already rising, the rate of rise will, at least in the
short term be increased. If we apply a negative feedback that decreases
the CO2 when the temperature is rising, the rate of rise will be reduced.



--
--
forging knowledge