A few thoughts from the trenches ...

tuttle@lxt.com (Mark Tuttle)
Reply-To: cg@cs.umn.edu
Date: Tue, 4 Feb 92 09:36:42 PST
From: tuttle@lxt.com (Mark Tuttle)
Message-id: <9202041736.AA03893@lxt.com>
To: ginsberg@t.stanford.edu, sowa@watson.ibm.com
Subject: A few thoughts from the trenches ...
Cc: cg@cs.umn.edu, interlingua@ISI.EDU, kr-advisory@ISI.EDU, srkb@ISI.EDU

As a potential (almost certain) consumer of knowledge exchange/representation
standards, in a distributed knowledge base building effort sponsored by the
National Library of Medicine (part of NIH), I would like to add
my voice to John Sowa's suggestions.

My colleagues and I suffer from the problems with the SQL standard,
problems which are liable to get worse, as a) the group responsible
tries to fix the problems with the original standard, and b) the
group responsible tries to respond to user needs for enhancements.
I believe the database community will be "paying" for SQL for years to
come.  I saw all this happen, and John is absolutely right.  Industry's
desire for a standard, removed the opportunity for the kind of meritocracy
we all take for granted.  When those who knew about the foundations of
query languages were ignored, they took their ball and went home.  It was
a BIG mistake.  Very big.

After thinking about the standards issue since the Parajo Dunes
Meeting of last Spring, I have come to the conclusion that
John Sowa's position is correct, namely that we need a logic
based standard.  Whatever shortcomings it may have vis a vis
anticipated non-monotonic reasoning engines, or whatever, should, in my
opinion, be handled at another level of abstraction, unless someone can
come up with a graceful, practical all-encompassing proposal now.
Alternatively, there could be one standard which standardized on
syntax only, and another, which uses the semantics of logic.
But, in my opinion, these are DETAILS.  I know some of you live
and die these details, but, believe me, you won't have a chance to get
to your details in a highly leveraged way, if we don't participate
and start this thing off on the right track (in general).

I cannot emphasize how important I believe it is to take a shot at a
useful implementable standard, and then make it a success.  In the
appropriate jargon, this group needs to 1) aim, 2) shoot, and 3) hit
the target.  If we can do that, we would have enough credibility to
propose another standard later.

David Patterson, Chairman of Computer Science at Berkeley, in remarks
to a computer science alumni gathering, observed that he saw a major
shift in computer science in the '90s -- away from "introspection" as the
chief methodology for research, toward solving large, important problems,
with whatever group of talent was required.  This places major burdens
on a field which has thrived on introspection, to learn a new set of
skills.  The payoff is that we can get to a position where we're
taken more seriously as effective problem solvers.  In my opinion,
this will necessarily lead to the emergence of a whole new sequence
of difficult technical problems, e.g. how DOES one organize, and
support hundreds of people into a knowledge building effort?

If I were running this thing, I would get the best minds in the field
and lock them in a room, until a progression of standards was agreed
on that could be evaluated by the user community.  In my view, this would
be immeasurably better than the alternative, which is to stay above
the fray.

A voice from the trenches,
-- Mark Tuttle, VP
Lexical Technology