> From: [log in to unmask] (Stu Weibel) > Date: Tue, 3 Dec 1996
> 
> Who else but the searcher can we really be thinking of?

I'm surprised to hear such a question.  Wasn't a reason for the Core
(our "small, essential set") the idea that we needed to be easy on
metadata providers?  From the beginning, we've talked about making
things easy on them:  metadata, we said, was going to be created and
maintained by authors, secretaries, and others who have no training in
cataloging.  Element names should be intuitive, encoding syntaxes should
not be awkward, and the entire DC should be simple enough that "the spec
shouldn't register on a bathroom scale" or people wouldn't use it.

> The whole > point is to facilitate discovery on the part of searchers.

And isn't the flip side of discovery the need of an author to publish
and be read, or the need of a vendor to advertize and be patronized?
Discovery is a complex end-to-end, provider-to-consumer process.

> To focus on the short term behavior of this indexing system or that
> indexing system is not productive in the long term.

This platitude doesn't help (it's always true that "to focus on the
short term behavior of <...> is not productive in the long term").
In one sense the whole point of the Dublin Core is short term; it
was to make a start -- to make a dent, to begin chipping away -- at
the rather complex problem of resource description.  While I'm behind
this sentiment, we do need to think long term about where the DC will
fit into more future general-purpose metadata (eg, the User Guide talks
about a Resource Metadata Vocabulary).

Anyway, I'd say we have a mix of long- and short-term goals.
But then, it doesn't much matter what I say, or what the next five
respondants say, since we haven't been explicit about our goals.
It's been fun beating heads against this tree with you all again.
Same time next month...?

> To the extent that the indexing community has useful insights about
> discovery (and they do), we should listen carefully.  That doesn't make
> them our audience, it makes them our partners.

If so, why aren't the Indexers more in evidence at our meetings and on
the list? Until they're really our partners, we have no clout with them.
Without the likes of Altavista, Lycos, and Infoseek, we probably wouldn't
have had these meetings for a few years to come.  Our work is of little
consequence to them now (even if they don't know, as some of us think
we do, what's "good" for them), and if we fail to understand their needs
we can expect them to ignore us.

Have we as a group done our basic homework, such as looking at what
elements the Indexers are using today?  Take, for example, your typical
internet search engine output of today, happily thriving, by the way,
without benefit of the DC; do you notice any elements in the search
results that aren't in the Dublin Core?  I do:  "Size", "Score",
"Search Stats".  These are not search elements, you say (well, Size
might be useful in restricting a search).  So what?  They still assist
resource discovery, in what I call the Selection phase.  We can declare
this Selection phase out of scope, but before we have a clear statment
of scope how do we know what we're collectively talking about?

> If we are doing something useful, then some or another indexing system
> will use this information to good effect, and the advantage derived
> from higher quality description will accrue to both the searchers and
> the services that provide it.
>  
> If we indulge in excessive complexity that renders our vision
> undeployable, then we will be relegated to the scrap heap.

These platitudes don't help.  Complexity is something we're stuck with.
Maybe simple metadata isn't as simple as we thought.  So we deal with it.
First we establish goals, audiences, and terminology.  They give us the
strong barriers we need to cordon off the simple corner of the problem
that we want to solve.

There's an enormous amount of good will, high energy, and creative ideas
that have been directed at this DC effort:  Stu, Jon, Paul, Cliff, et al
have all made excellent contributions, but there's no real foundation
to attach them to.  This wastes time, increases confusion, causes
burnout, and threatens our credibility.

We don't need a Formal Engineering Specification for the Dublin Core.
At a minimum, we need one spec with enough rigor and coherence to get
a passing mark in a University level writing and composition course.
I would be thrilled if we could achieve just that.

-John

> We are
> trying to find the right balance, and that means that the experimental
> work such as Jon and his cohorts are doing is enormously valuable.  So
> too, the reflections of Sigfrid, informed as they are by real world
> expectations of actual patrons.
> 
> Now, Sigfrid... what's this about rocking the boat?  I thought we were
> still laying the keel?
> 
> stu
>