RE: Issue 30 (Was: RE: Getting HTML5 to Recommendation in 2014)

Sam Ruby wrote:
> 
> In fact, there is also a point that I would like clarification on.  I
> would like to know if longdesc is only ever intended to be used in
> controlled educational environments with significant copyright
> restrictions and for that usage universal adoption by mainstream
> browsers is not a requirement?
> 

If I may:

I think we have unquestionably demonstrated that there is an immediate need
TODAY for a mechanism that addresses the need for longer textual
descriptions in the EDU sector. 

However, I will also point out that Geoff Freed of NCAM referenced a US
Federally funded initiative that is seeking to improve the overall corpus of
Longer Textual descriptions on the web, without specifically targeting that
to one unique vertical market. 

We currently have one publishing vertical that is signally an urgent need,
but that does not exclude others: Museums, government sector, health sector,
etc.; content producers publishing content that is not as ephemeral as other
popular web content.

In fact, we have collected evidence
(http://www.d.umn.edu/~lcarlson/research/ld.html#wild) from the Governments
of Canada, Australia, South Korea, Japan, the States of California and
Massachusetts, and more who are also using longdesc in their current
production workflow. We have identified other verticals as well, including
health (CDC, Substance Abuse and Mental Health Services Administration
(SAMHA), National Cancer Institute, Hipocampo), Art, Libraries or Museums
(Dayton Art Institute, Yorkshire and Humberside Books, Santa Barbara Public
Library), and Industry (IBM, the Linux foundation) so to suggest that this
is only for content of an educational nature is to color the conclusion in
an unhelpful way. 

Do all sites need @longdesc? Probably not. In Steve's raw data based on the
top 10,000 web sites home pages I noted with an ironic smile how many of
them were "adult" sites - they draw a *lot* of traffic but do we really need
a long description of "Suzi Q"? I suspect not. (Note, I make no value
judgment on the content of those sites - they have every much the right to
use the public internet as you or I, or the University of Victoria, the
Australian Government or CSS Squirrel - sites who are all producing
@longdesc content today.)

> Or is there a universal need for "long textual descriptions" that is
> not
> currently being met? 

I would suggest that this is a truer statement. The reasons why this need is
not being met likely can trace its roots to a number of reasons, but suffice
to say that one way of classifying the current situation is to reference the
movie "The Field Of Dreams" - there is little demand because there is little
supply, there is little supply because there is little demand.

This is a Catch 22 problem, and I would further suggest that the first thing
we need is a means to create supply, and easy access to that supply, at
which point the demand will likely increase. (ref: Field of Dreams, Henry
Ford and the Model T) Will it see massive adoption on the Top 10,000
homepages? That question has no relevance - many of those pages may never
need to use @longdesc.


> If so, what changes are required in order to get
> browser vendors to sign on?

A workable technical solution that meets the use-cases presented. There is
no argument that the requirement for longer textual descriptions will always
reside in the edge-case bucket, but when that edge case emerges, it is
critical that we have a solution to meet that need.

 
> Another possible answer is that this is indeed a market that long
> descriptions (by whatever the attribute is named) is a requirement for.
>   In which case, we need to take this data very seriously first the TF
> and ultimately the HTML WG as a whole will need to determine what
> corrective course corrections is needed.
> 
> And I will note that these answers are not necessarily mutually
> exclusive.  Perhaps we need two separate attributes.

I fear that 2 separate attributes "forever" would be overly complex, and a
burden to authors. 

I suspect that maintaining the status quo with @longdesc today, while
working on a superior implementation of a method to deliver longer textual
descriptions would prove more fruitful for all concerned. That effort could
follow one of 2 paths: improve @longdesc while retaining the attribute, or
work on a *future* replacement with a different name. If the latter, then a
managed path forward would also be a requirement; ripping and replacing will
have significant negative consequences on those examples of conformant
@longdesc that it would be counter-productive to do so.

I think that if you actually queried the majority of those involved with
this debate that you would find a broader support for a 2-phased approach
such as this, but I state that without actually asking others, and so only
offer that as personal opinion.

JF

Received on Thursday, 20 September 2012 22:35:09 UTC