W3C home > Mailing lists > Public > wai-xtech@w3.org > July 2007

managing alternatives

From: Al Gilman <Alfred.S.Gilman@IEEE.org>
Date: Sat, 14 Jul 2007 12:28:33 -0400
Message-Id: <p0611041fc2be9e82c96c@[]>
To: wai-xtech@w3.org

** intro

We have recurring problems concerning interaction design and the supporting
markup for that interaction design where some content is not universally
accessible and some alternate content is available.  How do we associate
the alternatives?  How do we manage the choices?

There are various places that this comes up:

a) the future of @longdesc in HTML5

b) accessible presentation of selectable-modality CAPTCHA tests

c) etc.

There are a variety of technologies that exist in some level of deployment
that do some of this:

- html:object

- smil:switch

- ssml:audio

One suggestion has been that we add an @equivalent attribute to the
WAI-ARIA vocabulary.

Even if we don't do that in WAI-ARIA 1.0, and even if we don't have
a fully-cooked proposal for what HTML5 should do in this area, we need
to communicate to HTML5 that we have this problem, or what our
goals would be for access under these circumstances. So I think we
need to talk about it.

** some discussion

* a linear fallback sequence is not in general the right algorithm 
for universal access


The author does not know a precedence order for what will best meet
the user's needs.  To best meet the needs of the widest variety of users,
all options must be available for selection by the user and/or the user's
decision rules.


In the past PFWG (and sometimes UAAG) has discussed with Voice
Browser, SMIL and the (then) HTML WG that content could be formatted
as a linear fallback sequence (nested if/then decision constructs implied)
*if* this processing were given 'default' or 'hint' semantics and the
client *may* treat the collection of all alternatives as if a
smil:par collection where any or several may be selected for

* the standards for equivalence are problematically vague.

In the case of highly task-oriented content, such as a CAPTCHA text
to try to rule out robot access to account creation at a service site,
it is clear what equivalent facilitation means -- you are not blocked from
creating an account.


In the case of passive content, the criteria for when another content
fragment should be accorded status as 'equivalent' is much more

In the educational-technology arena, there are things like 'learning
objectives' that make clear what 'equivalent' is supposed to mean;
"this learning object is supposed to enable learning objective (identifier)."
If two learning objects with different delivery-context dependencies
share that property, then it is appropriate to consider them substitutable
in tailoring homework to students' individual needs.


* there needs to be a well-understood way to return from a sidetrack,
if one is ever going to take such an alternative path in the dialog, in the
browsing experience.  You also need to know up front that there are
alternatives available.

* combo boxes are an existence proof for practice that works.

Here the alternatives are input alternatives, not display alternatives.
But the alternatives are clustered tightly enough, and the pattern is
used enough, so that users become acculturated to the pattern.

Received on Saturday, 14 July 2007 16:28:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:51:33 UTC