On Tue, 30 Dec 2014 15:54:02 -0600, F. George McDuffee
wrote:

On Tue, 30 Dec 2014 14:55:45 -0500, Ed Huntress
wrote:

snip
You're kidding yourself. Finding the things that "affect" the metric
is part of how they build models. It involves a lot of statistical
testing and regression analysis.
/snip

Old methodology called EvOp [Evolutionary Operations].

http://tinyurl.com/nffwfkz


You do not have to understand the process. Change a
variable, for example interest rates, and see what happens.
If the output moves in the "right" direction make more of
the change. If the output moves in the "wrong" direction
reverse the change.

Accurate record keeping is critical.

That'sw great! And on whom, or on what economy, do you conduct the
experiments? And how do you make everything else stand still, so that
you know what you're measuring?

In the kind of predictive modeling we've been talking about, without
an extensive program of regression analysis, you can't separate one
cause from another.


It is criminal that we allow the "leadership" to sabotage
the economy everytime things start to improve for the
average citizen. Things are getting better? Time for a war
or recession.

I doubt if any of them have much to do with it. In an economy like
ours, there are too many variables, too few unequivocal
cause-and-effect relationships, and too many confounding behaviors of
masses of people. Oh, and here are too many ideology-driven theories,
most of which assume that humans are uniform, rational robots. But
each theory assumes that we robots have different goals and desires.

--
Ed Huntress