[From Bill Powers (2007.11.20.0130 MST)]
Friedman talks about negative feedback, but clearly knows nothing about
loop gain. His negative feedback is a gentle low-gain affair that nudges
things politely toward equilibrium, rather than a robust high-gain
process that grabs the controlled variable, slams it into place, and
holds it there as if it’s nailed down.
I’ve just realized that his examples are full of controlled variables,
but they aren’t economic variables. For example, he speaks of the way
supermarket lines tend to adjust so they’re the same length. This is
because people don’t like to wait longer than they have to, so they
choose the shortest line, thereby lengthening it. Not liking to wait is
not an economic variable, it’s just an ordinary individual controlled
variable called “waiting time” with a reference level set to a
low value. The consequences of people doing this are economic, in that
the lines rebalance and become equal over time. That’s because the
actions taken to control the variable affect the environment and change
it, altering the environment that other control systems see. Latecomers
to the lines have no chance of reducing their waiting time because the
lines have all become equal. In fact they necessarily make the lines
unequal. When there’s only one shortest line left, the next person makes
them all equal, but after that …
So the social economic phenomena are side-effects of individual control
processes that create interactions among the control systems as various
systems seek whatever perceptions they want to experience. Economic
models need to contain individual control systems, if only representative
ones standing for collections of individuals with similar goals. Economic
discussions always do contain individual control systems, but they are
covert. They need to be made explicit.
The hidden model I’m looking for in Friedman’s book is going to be a
control-system model that he is using without being aware of it. When we
figure out what the people in his model are controlling, all the various
phenomena he uses as examples will fall out of the model. He’s looking
for economic laws by examining the interactions visible in the world
shared by all the control systems. He doesn’t realize that the real laws
are the ones operating inside the individuals. Bill Williams used to
pronounce the word “individual” with (literally) a sneer, as if
it were a heretical term that all economists with any sense (especially
his graduate advisor who he referred to reverentially as “The
Mentor”) eschew. But the real economic laws are consequences of the
way individuals control their experiences. The properties of a particular
economic system (laws and customs) determine how an individual
must act to control anything, but the perceptions and reference signals
inside the individuals determine what and how much is
created by behavior.
This is why Friedman’s arguments about “enough” are empty. He
thinks he has to get rid of that concept to make his model work, and he’s
right: if people spend less than all of their money, the interactions
that create conflicts mostly disappear. It’s only when some variable hits
a limit that actual conflict grows out of interactions, and the amount of
money available for an individual to spend is a limit. When there’s a
conflict, only one side, if either, can achieve its goals. Then you get
competition and all the rest of it. Otherwise people just slip around
each other and go on controlling for what they want without keeping
anyone else from doing the same. Friedman seems somewhat aware that such
an arrangement would screw up his system; at least he keeps coming up
with supposed reasons why it can’t happen.
The picture is gradually coming into focus, but it’s not there yet. I
wouldn’t mind if someone else worked it all out to the end before I get
there. Of course anyone who has too much of an axe to grind will not get
there first. It’s best just to let it come out the way it comes out, and
not care too much.
Best,
Bill P.