COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BDR25\M1BDI25\M2NGR30\M3XMAS25\M4NGB25\.
C00007 ENDMK
C⊗;
\\M0BDR25;\M1BDI25;\M2NGR30;\M3XMAS25;\M4NGB25;\.
\F2\CARTIFICIAL INTELLIGENCE LABORATORY
\CCOMPUTER SCIENCE DEPARTMENT
\CSTANFORD UNIVERSITY
\CSTANFORD, CALIFORNIA 94305
\F0

\CJune 5, 1973



Dr. Bertram Raphael
Stanford Research Institute
333 Ravenswood Avenue
Menlo Park, Ca.  94025

Dear Bert:

\J	Here is my report on \F1Beyond Omnipotent Robots\F0 by Gary G. Hendrix.

	The paper is a good one and should be published, because it raises
problems whose solution is vital for artificial intelligence.  I have one
major suggestion, several minor comments, and a general remark.

	1. The major suggestion is that the title be changed.  \F1Beyond
Omnipotent Robots\F0 suggested to me that it discussed what would happen
to humanity after omnipotent robots were achieved.  I felt I could postpone
thinking about that for a while, and this partially accounts for the lateness of
this referee's report.  I suggest a title like \F1A robot representation of
simultaneous actions and continuous processes\F0.

	The present title is inaccurate as well as misleading since the STRIPS
robots are not omnipotent even though they are the only actors in the situation.

	2. p. 8. The plural of \F1continuum\F0 is \F1continua\F0.  I don't insist
on correct Latin plurals if the editors don't.  p.11. The word \F1schemata\F0
is the plural of \F1schema\F0.  Its singular use is objectionable.
p.17. \F1discreet\F0 means \F1does not tell secrets\F0.  \F1discrete\F0 is the
opposite of \F1continuous\F0.

	3. The weakness of the formalism given in the paper is that it is
inadequate for the expression of partial information about processes, e.g.
humans can still deal with a situation in which it is not known how fast the
bucket fills up.  In the terminology of McCarthy and Hayes, the formalism is
not epistemologically adequate even though it may be metaphysically adequate.
It shares this weakness with STRIPS, although I think there will be even more
difficulty in expressing the information actually available to the robot
in the simultaneous continuous case than there is in the cases treated by
STRIPS.  

	In spite of this criticism, I favor publishing the paper, because
it comes closer to a metaphysically adequate formalism for simultaneous
continuous events than anything in the literature at present.  However,
the paper would be even more of a contribution if the author would discuss
representation of partial information about simultaneous
continuous processes.\.


					Sincerely yours,



					John McCarthy
					Professor of Computer Science
					Director, Artificial Intelligence Laboratory