COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	"Ambiguity tolerance" is the phrase we want with a slight bow to
C00005 ENDMK
C⊗;
"Ambiguity tolerance" is the phrase we want with a slight bow to
Dreyfus.

An ambiguity-tolerant representation language

Forms of ambiguity tolerance

Unambiguous notions of ambiguity tolerance

	After thinking about the problem discussed in this paper for a
few years, I came to realize that the term "ambiguity tolerance" introduced
in Dreyfus (1974) for one of the "things computers can't do" nicely
fits the objective of this paper.

	I am using Dreyfus's term in what I believe is approximately the
same English language sense he does.  Nevertheless,
I hope that this paper will not be
read with the main purpose of seeing whether it does one of the things
that Dreyfus claims computers can't do.  In the first place, Dreyfus
is not explicit enough about what he means by "ambiguity tolerance"
to provide a test.  Perhaps he thinks that such explicitness is contrary
to his doctrine.  Secondly, the forms of ambiguity tolerance we describe
are only a part of what needs to be done.

	I have decided to be explicit about exactly how much credit to
give Dreyfus.

	Communication is possible between people using terms on whose precise	
extension, they would not agree.  The simplest case is that people who don't
agree about whether whales are fish will not have any problems when one
says to the other "See the fish" pointing to a salmon.

	If we regard one part of a program as communicating with another,
they sometimes might have different meanings for a term and still communicate
unambiguously in particular situations, although I don't have any
examples realizing this.