COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	The first paragraph of the introduction proposes a more attractive (to me)
C00005 ENDMK
C⊗;
The first paragraph of the introduction proposes a more attractive (to me)
research program than does the abstract.  The problems posed in this
paragraph are fundamental to AI, and not much progress has been made.
Allen is also correct in pointing out the limitations of the state space
approach.  However, it seems to me that Allen's temporal reasoning
based on intervals covers only a small part of what is required to
solve these problems and so is excessively emphasized in the proposal.
Along this line it is somewhat disappointing that the example of
block stacking discussed in the proposal is one for which the state
space approach works all right.  When Allen and his students focus
their attention on examples for which the state space approach
doesn't work, I believe they will find that interval-based temporal
reasoning doesn't by itself solve the problem.

Allen is also correct in proposing a formalism that distinguishes
between what will happen and what might happen.  The examples of
the plan providing for a contingency and a the Little Nell problem
are good vehicles for the work.

In spite of my reservations about the application of interval-based
temporal logic, I consider this an excellent proposal.  Allen combines
theoretical work and implementation in an appropriate and successful
way.