Social interactions

[From Bill Powers (930415.1200 MDT)]

(methinks he doth apologize too much)
   McPhail (930415.0913) --

Clark carries on from where I left off. If we begin with the
concept of purposive actors each aware of the world and
controlling it from an individual point of view, we can re-enter
the world of sociology with a change not so much of language as
of meanings. This is what Clark is doing (and Little, apparently
-- thanks, Greg, for that citation, and who is going to sign up

A long time ago we had a discussion to the effect that social
systems are not control systems; only the people in a society are
control systems. But this doesn't mean that there are no
regularities in the interactions among the individuals. It means
mainly that those interactions aren't confined to the sorts of
interactions we see among the control systems within a single

As Clark has found in exploring the crowd simulation, the
interactions that naturally arise among independent control
systems can be characterized quite clearly, and emerge from the
individual control behaviors in predictable ways. When
individuals decide to maintain a respectful distance from a guru,
each making that decision independently, and also each avoiding
collisions with other people, the result is a complex interaction
that resolves into a beautiful arc or ring of people around the
guru when the guru stops moving. No person in the gathering
intended that an arc or ring should result from this; that is
simply a consequence of each person controlling privately for
certain (similar) perceptions. But it's a regular consequence,
often observed, and it's therefore a genuine property of the
group, a truly social phenomenon.

Here's a thought experiment that someone might want to try.
Collect three people together, and tell them to stand so that the
other two people are an equal distance away to the left and the
right (each person facing midway between the other two). If each
person adopts this goal, independently of the others, the result
will be three people standing in an equilateral triangle. This
result emerges from the three control actions, not because of a
property of the individuals but because of a property of geometry
in Euclidian space. The world in which we all (apparently) live
is so organized that the only way for the stated condition to
exist for each person is for all three sides of the triangle to
have the same length. So the equilateral triangle emerges without
any person having intended to create such a figure. Anyone who
understands the properties of plane geometry AND the properties
of individual control systems could predict this outcome of
adopting the particular goal. I haven't even tried it (only two
people here). Someone want to check it out?

Carry this on a little further. Clark mentioned cooperation. Two
people cooperate to lift something that one person can't easily
lift. Suppose it's a long uniform thick pole. What goal could be
suggested to the two people that would be likely to end up with
one person at each end, supporting the pole level with the
ground? What comes to mind is "Make the pole be three feet off
the ground and parallel to it."

This doesn't suggest how the people are to move, lift, or stand.
If both people move toward the center of the pole they will
interfere with each other. If they move apart a small distance,
they'll be supporting the pole from the side, an awkward
position. The easiest position is at one end, and one would
predict that they would end up each supporting one end of the

So the final configuration is almost an inevitable consequence of
both people adopting this particular goal. They distribute
themselves for maximum cooperation and minimum effort without
ever being told to do so. This is rather like Ed Ford's example
using eight rubber bands, four knotted into a quadrilateral and
the other four extending out from the corners. The suggested
reference perception for each of the four people holding the
loose rubber bands is "Make a square." The square appears almost
before the echo of the words dies away. The size can't be
predicted, but the squareness can. Yet no one person can produce
squareness alone.

So in developing sociology out of PCT, we bring in properties of
the world in which all the people live. Given that people have
the goals they have and the means of control that they have, and
given that each one's actions affect the others and the physical
environment, there are certain outcomes that can be predicted --
not from PCT alone and not from properties of the world alone,
but from the combination. When all the people in the stadium look
at the exploding scoreboard, we know that each person's line of
sight will lie along a radius extending from the scoreboard to
that person's eyes. If we know the positions of two people
looking at the scoreboard, we can predict the angle that will
exist between their lines of sight. So the relationships between
the control actions of the people fall out of the properties of
the world, if we have modeled the world correctly.

Communication, as Clark brings it into the picture, can be
handled in the same way. If the two people who have lifted the
pole are to carry it to some particular destination, they both
must know the destination and agree on it. That's a necessary
condition, set by the properties of the control systems and by
the inelasticity of the pole. What are the means by which the two
people could align their goals? Somehow, something must pass
between them that has that result. There are many possibilities
-- a gesture with the head, a slight brief tug on the pole that
the other can feel, or a sentence traveling through the air.
Perhaps many gestures, tugs, or sentences will be required before
the goals are actually aligned and the two people set off in
parallel directions. But we can predict that if the pole actually
gets to a specific destination, the goals will have become
sufficiently aligned to prevent conflict.

Of course, as Clark suggests, there is another possibility; a
third party issuing instructions (assuming that the two laborers
have already agreed to convert those instructions into reference
signals). But even without knowing how the goals get aligned, we
can state that the objective will not be reached until they are
aligned. That's another statement of a social fact.

It seems to me that this path leads to a very different sort of
sociology than the one that arises merely from watching and
classifying people's actions and interactions. There are
fundamental principles involved, resting on basic concepts of
individual behavior and concepts of how the world works. We can
state the sorts of things we have to know about the world and the
people in order to predict interactive outcomes. We can begin to
form a picture of social interactions in which we see the
outcomes as _necessary_. We see the equilateral triangle as the
inevitable result of each person adopting a certain goal for how
the other two are to be perceived, and doing whatever is
necessary to achieve that goal-perception -- in this space-time



Bill P.