OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

 


 

   Re: [xml-dev] Google as Big Brother

[ Lists Home | Date Index | Thread Index ]

Joshua Allen wrote:
> Pure sophism:
> 
> 
>>>First, Google is not mandated as central by any standards body. There
>>>is no "uddi.org" that decides who gets into the search engine game or
>>>not.
> 
> 
> That's like saying that Windows development is not centralized because
> anyone can write an operating system.  Technically untrue, and
> irrelevant besides.

Fair enough, I clarified that Google is designed as a component of a 
decentralized system, with no special status in the architecture, which 
is not the case with DNS or UDDI.

>>>Second, Google has no information about a site (except synthesized
>>>data like PageRank) that you cannot get from the site itself.
> 
> 
> This is patently false.  Google adds tremendous value as an
> "introduction broker" by tracking information from users of the "google
> toolbar", random clickthrough statistics, and all sorts of text analysis
> (not just links) of cached copies of the data.  None of that information
> is "free" or "open" by any stretch of the imagination.

Text analysis is synthesized data and anyone can do the same analysis by 
crawlign the Web themselves. The rest is really not that valuable to the 
users of the search engine. If it was, AltaVista would not have been so 
soundly beaten to a pulp by Google itself (they would have had such a 
huge advantage in "clikthrough statistics"). The clickthrough 
information etc. is much more likely to be useful to marketers than to 
users.

>...
> You don't even need to use them at all to *do* the introduction.  This
> argument is irrelevant.

Google's purpose is doing introductions (as is UDDI's). I have no idea 
what you're saying here.

>>>Fourth, Google's way of addressing resources is by their URIs. That
>>>means that once I get an introduction through Google, I can introduce
>>>a third party without going through Google again. UDDI uses UUIDs and
>>>those UUIDs are specific to some particular registry.
> 
> 
> Wrong about UDDI, and irrelevant because it's the "introduction broker"
> or directory services that we're talking about, not the actual visit
> once the introduction is brokered.

I'm not talking about the visit. I'm talking about introducing _someone 
else_. If I (an end-user) want to introduce someone else to IBM (as 
described in a bunch of TModels) I have to refer to those TModels in the 
UDDI repository, because there is no URI that has the same information. 
The information isn't (usually) a copy of information on IBM's site: 
UDDI is _the_ canonical owner of the information (as DNS is the 
canonical owner of the name->IP mapping).

If IBM happens to also have an RDF description of itself that happens to 
have the same information, I could use that instead. But obviously 
that's outside of the UDDI system. It should be the reverse. IBM should 
own their canonical information and the UDDI registry should have a 
copy. The UDDI registry should offer pointers to the IBM-maintained 
information and registries should compete with each other based on 
sophistication of queries supported and performance.

  Paul Prescod





 

News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS