Re: 'escape' method

At 8:04 PM +0200 4/17/05, Charles McCathieNevile wrote:
>This strikes me as a detailed description of why WCAG asks for links 
>to be grouped. The original idea was that one could indeed simply 
>jump out of the group - UAAG had a requirement that a user could 
>navigate a document tree - so if there is a group of links in the 
>tree they form a node, and the user can climb upand lok at the next 
>or previous node instead. (Sorry to those who don't think in XML 
>trees naturally - this might sound a bit strange).
>
>Al, does it make sense that you are talking about how something like 
>that ought to work in order to be helpful?


* summary

1. yes

2. not enough to say "group links"

3. we are in this CG thread trying to coordinate what we tell authors 
and UA developers

** details

1. Yes, this is the about the expected or assistive behavior that the
structural markup enables.

It is interesting that you represent the idea

"that a user could [...] simply jump out of the group [of links]"

as "the original" idea.

I have long labored under the idea that what most people understand
is the behavior of the 'skip-nav' link which allows one to get past a
thicket of time-wasters, but only from one point before you have
dipped into the range to be skipped.

I felt I had to explain that the desired UA function has a slightly
broader behavior than what we have been able to afford with a
hyperlink.

The simple thing would be to ask for a User Agent replacement
for the skip-nav link.  I felt we had to explain the difference carefully
to upgrade to an 'escape' method that provides the function
of the skip-nav link *and more*.

2. I know that for those working on the WCAG at the time "group
[related] links" meant wrap link groups in some recognizable entity
markup that would isolate and classify the link group.  But that's
not what those words mean in natural English after washing away
a fairly strong HTML-centric context.  So I think that we need
fresh words to be clear what we are asking of the authors.

It's not enough to have the links appear one after another. The
concentration so formed should be analysed for its common rationale,
and an entity identified with "links for that shared purpose" be part
of the markup [or object model] to facilitiate getting to and away
from that collection.

The thinking in this pass over this topic has been less sanguine
about winning uptake for a specific element, such as say, use the
html4:map element always as the link-group container; and more
sanguine about winning uptake for a more indirect method using a
known metadata pattern, such as an html4:link element with a
technique-identified value of the 'rel' attribute, with its 'href'
attribute pointing to the ID of an author-ad-lib element type.

3. Synchronizing what we tell content developers and UA developers.

This means that a preferred strategy identifies the 'escape' method
for User Agents, as well as the 'entify link groups' method for
authors as partners in the strategy.

We don't develop author techniques based on vague potentials and
independently decide what we want the User Agents to provide as
user-option functionality.

I don't believe that in the past were very good about saying "this
markup practice depends on that user agent practice to deliver
value to the intended beneficiaries."  This discussion of markup and
methods for link groups is a pilot attempt to do better in that
regard.

In other words, even if 'escape' were "the original idea" as to
UA behavior in the WCAG group, I don't believe that this method
ever became part of a consensus record of the WCAG process
and particularly never became a dependency-obligation on the
UA group to get User Agents to support that method.

Al

>cheers
>
>Chaals
>
>On Fri, 15 Apr 2005 19:50:31 +0200, Al Gilman 
><Alfred.S.Gilman@IEEE.org> wrote:
>
>>
>>This expands on Jon's agenda item about what we are marking in the content
>>and the [range of] anticipated assistive processing including UA behaviors
>>that will be enabled by this markup.
>>
>>** introductory overview
>>
>>Here I want to discuss two relaxations on the 'conventional wisdom' 
>>of accessibility.
>>
>>a)  problem: from
>>
>>What it is we need to mark?  Groups of related links.
>>
>>to
>>
>>What it is we need to trap and mark up better?  Concentrations of 
>>under-motivated links.
>>
>>b) solution: from
>>
>>Navigation destinations with Table-of-Contents -appropriate hint or 
>>label or link text.
>>.. supported by list and go-to methods..
>>
>>to
>>
>>Additionally an 'escape' or "get me out of here" method that
>>- mimics the 'escapable structures' feature in DAISY books
>>http://www.niso.org/standards/resources/Z39-86-2002.html#Escape
>>- suffices to replace skip-nav links in the content
>>- is similar to 'seek' functions in radio tuners and media players.
>>
>>** discussion
>>
>>I call it 'escape' because it is available from anywhere within an 
>>escapable structure, and
>>not only from the beginning of that structure.  It moves you past 
>>some current context
>>that is of a class that the 'escape' method is proper to.  In DAISY 
>>this is things like
>>forms and tables.  I think that navbars would be another category 
>>that this would
>>be good for.
>>
>>Skip-nav can be achieved by using the 'escape' method when at the 
>>start of the masthead.
>>But we would have to look at what the actual algoritm is so it 
>>doesn't take a confusing
>>number of 'escapes' to get to the main content in that way.
>>
>>This solves a more general problem.
>>
>>The users who have trouble being lost in navbars will have a 
>>problem wherever there
>>is a high concentration of links that are under-motivated and 
>>under-structured.
>>The visual browsing experience allows rapid skimming of the scene, 
>>so the usability
>>is not broken by a shotgun blast of many links with loosely 
>>affiliated context cues
>>as to what the linked resources might add to the experience.
>>
>>The eyes-free user and the high-cost-per-input-symbol user both need better
>>structured ways to manage their review of the action opportunities 
>>represented
>>by the hyperlinks.
>>
>>Also, while the Table of Contents metaphor is very valuable in 
>>organizing our work,
>>navigating indirectly through the Table of Contents is by its 
>>indirection less likely
>>to be used.  In the talking book experience, the conventional wisdom is that
>>we need to serve users who will approach the content as a topic tree and use
>>the table of navigation, and also users who will play the book like 
>>a serial tape and
>>may need some skip or local navigation commands, but generally ignore the
>>contents tree.
>>
>>I have been thinking we should try to adapt this bimodal use wisdom 
>>from the book
>>domain to the web page domain.
>>
>>There are of course limitations to the portability of the precedent.  
>>Books are
>>still, by and large, bigger than web pages.  And they are created 
>>in a way that
>>makes the table of contents more something that reflects the author's mind
>>than the way web pages get built.  But anyway, giving some way to skip out
>>of regions of rough sledding is an alternate model of what the 
>>assistive function
>>is that I think we should consider.
>>
>>Of course in the DAISY context we have high-political-correctness people
>>(the alternate format edition producers) controlling the markup of escapable
>>sections.  So it makes sense to make the skip command move to the end
>>of a marked section.  On the web we are not so fortunate.  So skipping to
>>the first sign of lower link density is practial, if heuristic.
>>
>>What I am angling for here is that we take a reasonably low level of assumed
>>commitment from the authors, but still consider an 'escape' method that
>>would be applicable to a) forms and tables without inspecting further markup,
>>and b) containers with 'roles' we designate as skippable, including 
>>'navigation.'
>>
>>Al
>
>
>
>--
>Charles McCathieNevile                      Fundacion Sidar
>charles@sidar.org   +61 409 134 136    http://www.sidar.org

Received on Monday, 18 April 2005 13:48:36 UTC