Operator no-change impasses
---------------------------

If an operator no-change impasse occurs, there are three possible reasons:
(1) the pre-conditions of the operator were not satisfied; (2) the operator
is incompletely specified (i.e., it needs to be further instantiated); and
(3) there are no productions that directly apply the operator and it must be
applied in a subgoal in its own problem space.

For case (1), the appropriate response is to use the same problem space and
search for a state where the operator will apply *(operator subgoaling)*.
Case (2) requires *operator instantiation*, and case (3) requires {operator
application}.  For both cases (2) and (3), domain-dependent problem spaces
must be available to perform the necessary problem solving.  Because
domain-dependent knowledge is required for the last two cases, the first case
is the default action.  Acceptable and worst preferences are created in the
subgoal for the super problem space.  These can be overridden by an
acceptable preference for a different problem space.

The operator subgoaling strategy is described in detail in Section 
See Operator subgoaling on page See Operator subgoaling.