Re: ISSUE-30 counter-proposal

Ian Hickson, Sun, 14 Feb 2010 08:54:07 +0000 (UTC):
  [...]
> == Rationale ==
> 
> Several studies have been performed. They have shown that:
  [...]
> * Most users (more than 90%) don't want the interaction model that 
> longdesc="" implies. 
> [http://webaim.org/projects/screenreadersurvey2/#images]

You don't find basis in that survey for saying the above. And I said in 
November [1] that I fail to see how that survey undermines @longdesc.

Here is why:

The survey gives the impression to focus on _screen reader behaviour_ 
rather than on _technical solutions_: "If a long, detailed description 
of these images is available, how would you prefer to have it presented 
to you?".

The final words of the survey regarding @longdesc was thus phrased in a 
user agent behaviour description wording: «the option of placing the 
alternative on a separate page but having it announced by the screen 
reader, the current behavior of images with the longdesc attribute, was 
a very unpopular option».

Behind the phrase "very unpopular" is hidden the fact that 9.1% 
_preferred_ the behavior which the survey claims as being unique to how 
@longdesc is treated by user agents. The rest thus _preferred_ other 
methods. This doesn't mean that that they can't live with other 
options. Very few of us human beings prefer to only have one option 
available.

The thing is: @longdesc was never designed to be "the preferred 
method". It was meant to be _a_ method.

But here we also see how the survey mixes user agent behaviour and 
mark-up options together. Because, whether the long description is 
presented by the user agent as a link, or as a very long alt text, or 
as a "longdesc-ish" something or as some fourth and fift option - all 
this depends on more than the user agent behaviour.

For example,the most popular variant was "As text on the web page, 
immediately following the image - 28.4%". This is not a description of 
a UA behaviour, however, but of a coding practise. We must eventually 
assume - or hope - that those that answered the survey were able to 
think about the user experience when a text follows immediately after 
the image.

One also gets to wonder what the second most option refers to: "As 
optional text, available on the same page but only if I request it by 
following a link". This sounds nice. But what does it mean? How can it 
be made to work that way? Can't @londesc work that way? Is it not 
optional to read/listen to the long description which the @longdesc 
points to? Again: A user agent cannot impact whether the optional text 
_is_ on the page or not - so again we must assume that those that 
responded were able to think about the user experience they get when 
things works that way. Of course they are able to do that, but the 
question in the survey was nevertheless unclear and with ditto results.

The text of the survey explains that most users preferred to have the 
long description on the same page. However, when we take the "longdesc 
preferrers" and the "link preferrers" together, then 28.9% of the 
responders preferred that the long description rather was in another 
page than inside the same page!

Ian, are you going to put that into HTML5? That 30% of blind users 
_prefers_ that he long desc is placed outside the page itself?

(Of course, again I must point out that whether the long desc is in 
another page, has nothing to do with user agent behavior [except in a 
derived way].)

Speaking about the "longdesc preferres" and the "link preferres": it is 
difficult to understand the difference between "On a separate page, 
announced by and available to my screen reader" (which the survey 
presentation presents as  "the current behavior of images with the 
longdesc attribute") and "On a separate page, available by following a 
link". I mean: Isn't a link "announced by and available to my screen 
reader"? What is the difference between @longdesc and @href in that 
regard? Both a @href and a @longdesc contains a URL, and both of them 
must thus be presented to the user as a link. 

In conclusion: I find it hard to see how one can say anything about the 
"interaction model" of longdesc based that survey.

  [...]
> ARIA provides a number of alternative mechanisms that are currently not 
> poisoned by existing content and that fit better into the kind of 
> interaction model desired by users (according to the survey cited above). 

I fail to see that you have presented a interpretation of the Webaim 
survey results which supports this.

> For example, aria-describedby="" allows an image to be related to in-page 
> descriptive content.

@longdesec allows the same thing - it allows an image to be related to 
in-page descriptive content. 

And remember: 30% of the users in the Webaim survey _preferred_ the 
behaviour that user agents show when the long description is on another 
page. The good news with @longdesc is that as long as you link to the 
long description via @longdesc, then it will be presented to the user 
_as if_ the resource is on another page. 

[1] http://lists.w3.org/Archives/Public/public-html/2009Nov/0056

-- 
leif halvard sillli

Received on Monday, 15 February 2010 18:11:35 UTC