[a11y] Requiring alt Re: fear of "invisible metadata"

On Thu, 21 Jun 2007 18:31:53 +0200, Laura Carlson  
<laura.lee.carlson@gmail.com> wrote:

> On 6/18/07, Maciej Stachowiak <mjs@apple.com> wrote:
>> ALT should not be required either. It leads to pointless alt="" on
>> images that have no reasonable text equivalent, just to satisfy
>> conformance checkers. And that is actively harmful, because AT can't
>> tell the difference between a semantically null image and a
>> semantically meaningful image with no text alternative.
> ALT should be required. The single most important thing a person can
> do to make a web page accessible

*to blind people* - which is a small fraction of people with disabilities.  
Not to suggest that they are somehow less important than the relatively  
larger number of people with low but some functional vision, the smaller  
number of people who are deaf-blind, or people who are colour-blind,  
motor-impaired, suffer from cognitive impairments or have other issues in  
perception. Just to remind people that while accessibility for blind  
people is important, and is in many cases well-studied, it is far from the  
only aspect of accessibility and in terms of raw numbers not even very  

> is to include alternative text for
> images with alt attributes. "If there is no alt attribute at all
> assistive technologies are not able to ignore the non-text content."

Well, in practical terms, a user agent has to try and provide some help  
since this situation is a common accessibility problem, and it does it by  
more or less intelligent guesses.

> An alt attribute is not a label or description for the image...

Indeed. This is a very important point, and also bears on questions like  
whether longdesc should remain in HTML.

> Let CSS handle presentation
> and HTML provide the structure and semantically marked up content.
> Separate content and style.


> So my suggestion is to require alt for ALL images. But discourage the
> use of  eye candy images in HTML, instead recommend and promote the
> use of CSS for such purposes. Perhaps even consider _gracefully_
> degrading empty alt/eye candy in an orderly manner.

I agree with the conclusion. Requiring alt - and alt="" as appropriate,  
 from authors, is a good idea. And it agrees with WCAG (I don't always, but  
not having to argue across groups is a win :) ).

The Authoring Tool guidelines [1] say that *tools*, from Frontpage to  
Flickr, and my.opera to my custom made CMS, should not simply add alt=""  
or any other default text to correct the error of missing alt, since in  
fact the error condition is more useful than a bad alt text. I understand  
that some designers will not do this, instead putting meaningless alt text  
to pass some poorly-framed requirement they are working to. But I don't  
see another solution that makes sense of the web today and helps to  
improve accessibility in the future.

(Unless you are prepared to go down some kind of RDF-style collaborative  
annotation path. It seems that is still a while away).

Buried in this is a long debate about conformant content and  
accessibility, which I won't go into here, but which has raged in the  
accessibility community for many years. (stupid aside: I, of course, have  
the right answer, but so far haven't been able to convince the entire  
world of that, and I don't have the ability to just put it into a spec and  
claim victory, so I expect the debate to continue a while yet).



   Charles McCathieNevile, Opera Software: Standards Group
   hablo español  -  je parle français  -  jeg lærer norsk
chaals@opera.com    Catch up: Speed Dial   http://opera.com

Received on Thursday, 21 June 2007 17:56:32 UTC