Re: SUO: Ballot with 2 Questions
Mike,
Those of us who have been working in AI for more years
than we care to recount are frustrated by that point:
> Very few pieces of AI have been successful in the marketplace.
I'd be happy to write some flames about the marketplace,
research grants, and monopolies. But to get to the point,
Lenat and Co. have been trying for years to support commercial
businesses with their ontology, but they haven't succeeded.
And I'm not blaming Lenat or his colleagues, many of whom are
very very good at what they do. There may be other reasons.
Following are some possibilities:
1. The world doesn't need or want an ontology.
2. A good ontology isn't possible.
3. The methods used by Cyc aren't appropriate for
developing a good ontology.
4. There are still many missing tools that are
needed before an ontology can be connected to
the usual commercial software.
5. There is something wrong with the assumption that
an ontology is a prerequisite for intelligent systems.
I honestly don't know which of these (or any other)
possibilities should be blamed for Cyc's lack of commercial
success. But one thing I do feel strongly about: trying
to imitate Cyc on a small scale with "just" an SUO is even
less likely to be successful than Cyc has been.
That is why I have proposed motion #2 as a shift in
emphasis from what Cyc has been doing. I don't regard it
as competition with Cyc nor as an attempt to redo Cyc on
a small scale. It differs from Cyc in the following ways:
1. Take advantage of Cyc's experience by using OpenCyc
as a resource (along with the results by anybody and
everybody else who has been working on ontology).
2. Provide better tools (such as the IFF methodology
and a registry).
3. Enable those tools to support legacy applications
(which have been making lots of money for the people
who use them and develop them).
4. Emphasize modularity, extensibility, and flexibility.
Of course, Cyc could also adopt any tools or techniques
that the SUO may develop. That is fine, since I believe
that we should collaborate with them rather than compete.
But I also believe that we should emphasize the potential
for linking the SUO efforts to commercially successful
systems for the same reason why Dillinger robbed banks:
that's where the money is.
John Sowa