We know that there are two types of messages: descriptive, or
statements, and imperative, or commands. The combination
of the two creates a relation which is known as control.
A system C controls system S if C receives from S statements
and sends to S commands. Actually, this is a special case which
is predominant in well-organized cybernetic systems. In the general case,
the controlling system C also may directly alter the states of the
controlled system S. The difference between direct intervention and
sending a command is relative, though. It depends on the definition of
the systems involved.
Example: you drive a car. Here C is you, and S is your environment:
the car, the road, etc. You receive visual and audio messages from
the environment and vary the position of the car's wheels to keep
it on the road. This is a direct change of the state of the controlled
system. But you can consider your steering wheel as an information
channel, through which you send commands to the wheels proper.
Another example: remote control of a missile. Here you receive information
from the missile's sensors and send commands to it. But if you do not
wish, for some reason (although this is not really reasonable) to
separate the information channel from the missile and consider them
together as the controlled system S, then you exercize a direct
intervention.
The relation between a language and metalanguage, or theory and metatheory,
is also a control relation. Here S is the system that uses the language,
and C is the system that uses metalanguage. We create metalanguage
and metatheory in order to examine the works of the language and the theory.
The flow of descriptive information is from the language S to
the metalanguage C. The result of our examination of the language and
theory is classification and construction of more sentences
of the language and more theories -
this is alteration of the state of the language S.