Re: fear of "invisible metadata"

On Jun 18, 2007, at 10:40 PM, Ben Boyle wrote:
> The idea that the table summary should be accessible in visual media
> is good. It may seem obvious that any UA *could* access this
> attribute, but the HTML4.01 spec (and XHTML2) are quite clear that the
> intent behind table summaries is "rendering to non-visual media such
> as speech and Braille" [1][2]. So it was not intended to be used in
> visual media? Maybe I am reading the spec too literally.
>
> Many tables benefit from some extra explanation (i.e. a summary) about
> their structure. This can be useful to everyone, if it is not limited
> to "non-visual media". I believe this is what the WCAG Samurai refer
> to in their advice.
>
> But if use something outside the <table>, something other than
> table@summary, then that information isn't explicitly associated with
> the table (and it becomes more difficult for AT to make that
> association). But is it better to duplicate that information outside
> the <table> for visual media (WCAG discourage duplication with
> "summary should not duplicate the caption" [3]) or to provide nothing
> in the visual media? The idea that tables presented visually never
> require an additional summary is a bit narrow I think... not really
> taking into account the opportunities for improving information
> comprehension that the summary could offer.
>
> I think we should retain summary BUT remove the note about non-visual
> media in favour of adopting universal access. We should be encouraging
> its inclusion in visual rendering (I think it should support the same
> kind of styling, especially positioning, that <caption> does [4]).
>
> Unless there are reasons for keeping @summary limited to non-visual  
> media?
>
> [1] http://www.w3.org/TR/xhtml2/mod-tables.html#edef_tables_summary
> [2] http://www.w3.org/TR/html401/struct/tables.html#adef-summary
> [3] http://www.w3.org/TR/2007/WD-WCAG20-TECHS-20070517/ 
> Overview.html#H73
> [4] http://www.w3.org/TR/CSS21/tables.html#propdef-caption-side

In general, its good to provide semantic elements that are  
universally consumable. However, that is not always the case. For  
visual media (including tables), there are certain semantics that a  
visual user agent makes immediately available to a sighted user that  
it can't make available to visually impaired users. image@alt,  
image@longdesc, table@summary and table > summary are all ways of  
adding information that a sighted user already has access to. For  
example a picture of "a snowcapped mountain with a forest in the  
foreground and a brilliant orange sun setting to the right of the  
mountain peak." is not the type of description one would put in the  
caption/legend of the photo. It is already apparent to the sighted user.

So these inherently visual elements have associated elements for  
universal descriptions (caption, legend). However, when (one needs or  
wants to provide additional semantics to the non-sighted user, these  
other mechanisms exist. Used properly, the sighted user will not be  
interested in reading the long description (except to make changes or  
otherwise review it). We should try to provide mechanisms for  
universal access where possible. However, we should also continue the  
practice of providing other alternate mechanisms for improving  
accessibility.

Take care,
Rob

Received on Wednesday, 20 June 2007 23:59:07 UTC