what's a site, anyway?

to follow up on what John T. Whelan said:

> 	Speaking of Bobby, I have to express some confusion over
> the "If none of the web pages on your site contain
> accessibility errors" condition.  How does one define a "site"?
> My web pages fall into several self-contained units, but each
> of them lives on a larger site over which I have no control.
> Is the Utah relativity homepage
> <http://www.physics.utah.edu/research/relativity> only allowed
> to display the "Bobby Approved" icon if all of the pages on
> <http://www.physics.utah.edu/> pass muster?  That's impossible
> unless we get every member of the department to make their
> personal pages accessible.

1.  Policy issues like this question are just what we created the
Evaluation and Repair Interest Group to consider.  If there is a
lot of follow-up, you may want to take the conversation there.

2.  It is sooo great that the _one person_ in the unverse who
understands that the notion of a "site" is unnatural for the Web
is the keeper of the _relativity_ web presence!  Actually, TimBL
understands, too.  But many people don't.  Site-think is a
triumph of pre-existing culture with a strong undercurrent of
hierarchy over the flat, peer-heavy legacy from the Internet to
the Web.

3.  However, until somebody from Bobby tells you otherwise three
times [actually, they are most cooperative] to the contrary,
please proceed to apply the site-qualification rule as if
"site" is any collection of web pages which:

	a) evaluated as natural language, address a cohesive topic, and

	b) evaluated as hypertext fragments, implement a consistent
	policy of navigation-aid linking so that a visitor is unlikely
	to get lost, and can easily contact a person or business entity
	who will take responsibilty for the content of the pages.

Most people think of a site as a hierarchical realm in URL space,
but the above definition allowing general graphs of pages is more
appropriate because intentional, cohesive "web presences" can and
sometimes should be constructed which span DNS nodes and have
arbitrary shapes in URL space.  But if the user can't find
themselves when lost by truncating the URL of the current page,
you have to be scrupulous about your navigation aids.

Al

Received on Tuesday, 18 August 1998 12:48:08 UTC