Re: Report on ANSI X3H4 meeting

YOUR NAME HERE <hayes@sumex-aim.stanford.edu>
Date: Mon, 3 Feb 1992 13:22:26 PST
From: YOUR NAME HERE <hayes@sumex-aim.stanford.edu>
Subject: Re: Report on ANSI X3H4 meeting
To: sowa@watson.ibm.com
Cc: kr-advisory@ISI.EDU, srkb@ISI.EDU, interlingua@ISI.EDU, cg@cs.umn.edu,
        mzemanko@note2.nsf.gov, jwinkler@nasamail.nasa.gov,
        skperez@mcimail.com, jfulton@atc.boeing.com, mjones@atc.boeing.com
In-reply-to: Your message of Sun, 2 Feb 92 22:15:00 EST
Message-id: <MacMS.62098.2749.hayes@sumex-aim.stanford.edu>


John's helpful and informative message illustrated for me what it is that
still chills my blood when I hear talk of "standards" for Krep languages. Much
of this stuff is way out at the edges of conceptual research.  "Contexts" is a
vague idea which probably will separate into at least three ideas when looked
at carefully.  It has so far had one - that is ONE - PhD thesis devoted to it,
and has not been seriously used by anyone. (The CYC usage is essentially as a
heuristic device to allow 'knowledge-enterers' to avoid awkwardness, but the
deductive implications of 'microtheories' have not yet been explored and
different members of the project have different ideas about what they mean. 
For example, should all facts be in some microtheory? Do microtheories form a
lattice-like structure or should they be thought of rather as a 'covering' of
the knowledge base? Do they indicate deductive relevance? Do they encode
hidden assumptions, or represent a focus on some topic? Do they correspond to
temporally connected facts, and if so is there any notion of persistence
associated with them? You can get different answers on all these questions.)
And people are talking about incorporating it into a standard??  I know the
conclusion was that there are many unresolved issues concerning them, but they
shouldn't be even considered for 'standardisation': the idea should be a joke.


John tells us that standards are coming, but when this kind of thing happens
maybe the best thing responsible academic researchers could do is to criticise
the idea as strongly as possible, or at any rate point out when it becomes
nonsense. Sorted languages we understand pretty well, and maybe there some
standardisation makes sense, but standardising contexts before J. McCarthy,
their inventor, has published a paper on them is like weighing a ghost.

Pat Hayes

-------