COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	janler.com[w83,jmc]	Notes on "Modelling change - the frame problem" by Janlert
C00005 ENDMK
C⊗;
janler.com[w83,jmc]	Notes on "Modelling change - the frame problem" by Janlert

	This paper gives a philosopher's reaction to the situation
calculus, the frame problem and related problems.  He says, "I do not
claim that philosophy has all the answers, but I think that any
significant progress will hinge on our ability to acquire a
philosophical perspective".

	It seems to me that Janlert, like almost all philosophers,
doesn't see the value of formalisms with limited capability.
He also isn't quite fair in saying that Fikes and Nilsson thought
they had solved the frame problem or that McCarthy and Hayes
thought that situation calculus would solve all problems.  Our
respective formalisms were the best we could do at the time.

	A few comments relevant to specific pages:

p.7. He should mention that some entities needn't be depend on the
situation, e.g. we write  is-block(x)  rather than  is-block(x,s).

p.13. The point isn't so much that a detailed representation makes
the computation slow but rather that the facts required to use
the detailed representation can't be obtained.

p.15. The frame problem is indeed a case of the qualification problem.
The proposed relation between the qualification problem and the
philosophical notion of essential properties seems dubious.  A property,
e.g. that a tool is usable for its customary purpose, that is to
be taken as a defeasible presumption isn't the same as what philosophers
have called an essential property.

p.23. A proper treatement of STRIPS and the frame problem requires a
notion analogous to the physicist's or geometer's notion of a
co-ordinate system.

p.25. STRIPS can do the up and saucer problem just fine.