RE: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)

["Joshua Allen" <joshuaa@microsoft.com>]

> What is there to disagree with here?  I suppose you could say that

Not much really, that's the nature of straw men.  90% of what he says
could be said about "data" as well as "metadata", and the whole rant
reminds me of the bomb shelter crew who are convinced that one day a
computer glitch will bring the whole way of life to a screeching halt.
Or that John Poindexter's data mining program will go haywire and start
putting innocent libertarians in jail.  All of these problems; lying,
lazy data-entry clerks, etc. have existed since we started to store
information in computers, yet many enterprises are past terabytes of
data with no signs of slowing down.

In addition, the arguments are crap because people already store and
manage terabytes of metadata, and are doing just fine.  Information
about location-based services, document management libraries, PIMs, etc.

I can think of many more reasons that Doctorow cites that data might be
inaccurate.  I have seen many of these additional reasons manifest in
reality.  BFD.

> depressed. Conversely, I guess in TimBL's version of meta-utopia as
> opposed to the  META tag strawman that Doctorow demolishes,

I don't recall TimBL ever claiming that we would have a meta-utopia of
completely trustworthy metadata.  This, too, is a straw man.  Doctorow
spends so much time trying to convince you that metadata might be buggy,
as if anyone ever asserted that it wouldn't.

> Doctorow's argument because RSS feeds are an existence proof that
> useful metadata is practical.  I'm not sure which of the straw men
that

Well, this is exactly the point.  "Useful" and "infallible" are two
completely different things.  Again, this issue has nothing to do with
metadata.  On IRC you can never know who you are talking to with
certainty, and people *do* lie.  However, IRC can be very useful.  All
of life is about making assumptions under ambiguous conditions.  If you
can find one thing that is infallible in this human existence, then I
will concede the point.  Otherwise, just accept that life is full of
ambiguities and focus on what's useful.  There are countless "useful"
ways to harness metadata, regardless of ambiguity.

> On the other hand,  Doctorow's "screed" does call into question the
> WinFS vision, or am I missing something here?  To what extent does
> WinFS not presuppose honest, energetic, intelligent, and self-aware
> humans to create the metadata it will manage and query?

Completely irrelevant, IMO.  Think of WinFS this way -- you have lots of
things which you *already* capture metadata about.  Such as contact info
in your address book.  Or information about songs, or movie files.  You
*already* manage this metadata, and have been for a decade at least,
probably.  The world hasn't come crashing down.  

Now, all WinFS does is take all of that metadata and store it in a
consistent, unified, interoperable way.  So now you don't have to use
different APIs, different tools, etc. to get at the metadata.  You still
control whether or not you import/export someone else's metadata.  You
can't see my contacts if I don't let you; I can't see yours.  So it's
exactly the same as we have today.  However, the fact that the "wall"
around the metadata, and the "walls" between the different types of
metadata, have been removed; means that you have much greater potential
for interop and synergy.  It might seem like a very minor step, but
really the value of WinFS is that you have "liberated" your metadata
(well, a lot of it :-))from the clutches of proprietary formats and
APIs, and that is something I think is going to have a *huge* impact.
And can anyone *honestly* argue that "no, PIM data should be hidden in
as many different and incompatible data models as possible, because
metadata is crap!"??

Semantic web is similar vision, but on a global scale.  We recognize
that people are already storing and managing terabytes of metadata; so
what will happen when we "liberate" that metadata in a way that allows
repurposing, sharing, and synergy?  Presumably, there will be many new
applications that depend on metadata, more sharing, and more incentive
to collect (automated or manual) and maintain new metadata.