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

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

>> Google is to "the semantic web" as CompuServe was to "the web".

> words, what makes a SemWeb database more useful than the 
> technologies we are using now?  Where will it take us today?  
> What will we implement that we aren't already implementing?

Well, I chose the analaogy because it highlights what I think is the
critical disctinction.  CompuServe and others had hyperlinked content
prior to the WWW.  You could argue (and people *did* argue) that what
CompuServe provided years earlier was actually *superior* to what the
WWW offered initially.  There was a specific quality about WWW that
excited us in those early days, despite the relatively crude quality
compared to CompuServe, and which made it clear that CompuServe would
eventually fade.  It's the same quality that I think differentiates the
vision of "semantic web" from utilities like Google (or MSN Search for
that matter).

The critical quality is universal openness.  To get hypermedia published
on CompuServe or AOL, you had to strike a deal with the network.
Hyperlinks in AOL looked different from hyperlinks in CompuServe; one
system could not link to the other.  These systems relegated you to
status of sharecropper on the network's walled garden.

By the same token, if I want to get a new field of metadata published in
Google's index, it's even more difficult.  I don't even know who to pay.
The engine indexes things like pagerank, related pages, keywords.  But
if I want any other information in there, I have to get a job as a
janitor at Google's offices and hack the source code.  Now, if I build
my own engine for storing additional metadata like annotations, ratings,
etc., I have to build adapters (and probably violate some EULA) to get
that to integrate with Google or MSN Search.  These systems are all
walled gardens.

I'm not saying that the "walled gardens" have no value.  In fact, they
are extraordinarily valuable.  But they are sitting in exactly the same
position CompuServe was sitting 10 years ago.  People often focus on the
current value, and argue that "the landlord is a benificient landlord,
so what possible benefit would I get by abandoning the walled garden for
some university-produced science project?".  But I remember these same
arguments about CompuServe; I'll concede that Google is very valuable,
but I think that universally open metadata network will be immeasurably
more valuable.

When we talk about "semantic web", we are talking about the vision of
universally open metadata, NOT about walled gardens like Google.  It
doesn't matter whether or not you believe the vision to be attainable or
not; the vision is what it is, and it would be mischaracterizing the
vision to say that a walled garden somehow fulfills that vision.  Saying
"the semantic Web is already here, and it is called Google" is like
saying "The WWW is already here, and it is called CompuServe".  Both
statements are incorrect, in that they assume away the fundamental thing
that characterizes the web and semantic web; their universality.