RE: [xml-dev] Semantic Web permathread, iteration n+1 (was Re: [xml-dev] InfoWorld agrees with Elliote Rusty Harold)

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

> This is the same *data* people see and use. It is *not* 
> metadata. The semantic web assumes two separate layers of 

Google is incredibly dependent on several forms of metadata:
* Every time you publish a web page that contains a hyperlink to another
page, you are asserting that "the target page is interesting to me".
That is metadata in every sense of the word.
* When you click through a Google search result, you are asserting that
"this search result was more relevant than the others".  Google captures
this metadata randomly.
* When you browse pages, you are asserting that "this page is
interesting enough to spend time on".  If you have Google toolbar
installed, and the appropriate opt-in selected, Google captures this
metadata
* When a certain number of pages around the world link to a particular
page, that page is given a higher "page rank".  This is metadata in
every sense of the word.

> Sites can and will lie with their metadata. Google, by 

You seem to be assuming that metadata is only something which is
published by a page author about his or her own page.  Metadata about
content which you own is no more interesting than hyperlinks to content
which you own.  The point of semantic web is to permit anyone to publish
metadata about anyone else, the same as the point of the web was to
allow anyone to hyperlink to anyone else.  In fact, you can think of the
WWW as a version if the web which had only one term in the vocabulary:
href.

You also seem to be assuming that metadata is something which must be
published by a human.  Again, I would argue that some of the most
interesting metadata available is implicit, and needs no human
interaction to maintain.