Re: User Testing footage of header/id combinations, @summary and @longdesc for HTML5 WG

On Sep 14, 2007, at 6:23 AM, Robert Burns wrote:
> Hi Eric,
>
> On Sep 14, 2007, at 7:25 AM, Eric Eggert wrote:
>
>> Tomas Caspers, an German accessibility expert, wanted me to share  
>> his thoughts about the @longdesc topic with you. He's referring to  
>> this WHAT WG blog entry: http://blog.whatwg.org/the-longdesc-lottery
>>
>> Here are Tomas’ thoughts:
>>
>    [ A few anecdotes about the 'longdesc' attribute ....]
>>
>> Conclusion: I'm sorry folks, but I'm afraid you will have to come  
>> up with something else other than longdesc.
>
> I don't really understand how anyone can go from Thomas's few  
> interesting anecdotes to that conclusion. The longdesc attribute is  
> there to provide a long description of an image embedded with an  
> IMG element. In many cases it may not be useful at all to provide a  
> long description. However, for authors that do provide long  
> descriptions it is an indispensable attribute. For the users who  
> make use of the attribute and its referenced document fragment it  
> undoubtedly enhances their user experience.

I apologize if I've missed a vital part of the conversation (and I'm  
certainly no expert on accessibility technologies), but it seems to  
me that Tomas is saying that there are no "users who make use of the  
[longdesc] attribute".

> So I have to ask the question. How does Thomas go from his  
> anecdotes to the conclusion that we folks must come up with  
> something else other than lohgdesc? What would that something else  
> do? Make the blind see? I really don't see what he's saying at all  
> in relation to the work of this WG. I don't even understand how  
> anything Thomas wrote could be related to a decision about whether  
> to include a longdesc attribute or not in HTML5.

Sure, its anecdotal evidence, but if its true, we're talking about  
thousands of sites being formally evaluated for their accessibility  
and none of the evaluations involves a longdesc.

Tomas ancedotes are relevant to this WG, because it's evidence  
supporting the assertion that longdesc is rarely used, and when it  
is, its rarely used usefully. This useage has lead to users not  
making use of longdesc attributes.

There are plenty of alternatives between "using longdesc" and "making  
the blind see". There are other approaches to identifying the content  
of images on pages, which may provide a better experience for those  
using ATs.

-ryan

Received on Friday, 14 September 2007 18:30:52 UTC