W3C home > Mailing lists > Public > public-html@w3.org > April 2009

RE: Example canvas element use - accessibility concerns

From: Ian Hickson <ian@hixie.ch>
Date: Sat, 25 Apr 2009 22:17:15 +0000 (UTC)
To: "John Foliot - WATS.ca" <foliot@wats.ca>
Cc: 'HTML WG' <public-html@w3.org>, 'W3C WAI-XTECH' <wai-xtech@w3.org>
Message-ID: <Pine.LNX.4.62.0904252158100.10370@hixie.dreamhostps.com>
On Sat, 21 Feb 2009, John Foliot - WATS.ca wrote:
> > 
> > If the image is actually a _replacement_ for text that otherwise 
> > existed, why would the user care that the author had made the 
> > stylistic choice to use an image instead? I don't understand.
> > 
> > Specifically, the example in the spec is talking about an author who 
> > has something like this:
> > 
> >   <p>In the common case, the data handled by the tokenization stage
> >   comes from the network, but it can also come from script.</p>
> >   <p>The network passes data to the Tokeniser stage, which
> >   passes data to the Tree Construction stage. From there, data goes
> >   to both the DOM and to Script Execution. Script Execution is
> >   linked to the DOM, and, using document.write(), passes data to
> >   the Tokeniser.</p>
> > 
> > ...and decides to replace the text with an image. Why should there be 
> > _any_ difference for the non-visual user? What the spec proposes:
> > 
> >   <p>In the common case, the data handled by the tokenization stage
> >   comes from the network, but it can also come from script.</p>
> >   <p><img src="images/parsing-model-overview.png" alt="The network
> >   passes data to the Tokeniser stage, which passes data to the Tree
> >   Construction stage. From there, data goes to both the DOM and to
> >   Script Execution. Script Execution is linked to the DOM, and, using
> >   document.write(), passes data to the Tokeniser."></p>
> > 
> > ...seems like exactly what the user would want. Why should the user 
> > who doesn't have access to images be affected here?
> 
> OK, fair questions.  The problem however is that you are making a 
> judgment call of what the 'user' might want based upon your subjective 
> point of view, which is based upon your subjective interpretation of the 
> author's intent. How do you know these things?

In the case of the example above, I'm both the author and the reader.


> You yourself are not a blind user, so how do you know what any blind 
> user might want, never-mind all blind users?

This isn't only about blind users. It's about users who can't use the 
image. I am such a user on a regular basis, as I use various text-mode 
browsers.

(Obviously, though, each user may have different desires.)


> Thing is, the author didn't provide text, he provided an illustration
> (parsing-model-overview.png).

Actually in this particular case the author wrote the text then wrote the 
illustration to replace it. (I know, it was me.)


> So the image then is more likely a flow chart, or a Venn diagram or 
> something similar to that - and that is legitimate, but introduces an 
> aspect beyond simple text representation as it is a more complex 
> visualization.

How so?

The image is the one in the spec:

   http://www.whatwg.org/specs/web-apps/current-work/images/parsing-model-overview.png

It really doesn't convey any more than the text given above.


> So, to solve the 'problem', I would likely go with:
> 
>    <p>In the common case, the data handled by the tokenization stage
>    comes from the network, but it can also come from script.</p>
>    <img src="images/parsing-model-overview.png" alt="flow chart illustrating
>    the parsing model overview" longdesc="parsing_model_overview.html">
>    <p>The network passes data to the Tokeniser stage, which
>    passes data to the Tree Construction stage. From there, data goes
>    to both the DOM and to Script Execution. Script Execution is
>    linked to the DOM, and, using document.write(), passes data to
>    the Tokeniser.</p>

As a user who regularly has to view the Web without images, I assure you 
that I would find this significantly less useful and more annoying than 
the second example above.

As a text-only reader, I really *do not care* that there is an image here. 
The image doesn't add anything to the discussion; calling attention to it 
feels like the author laughing at me for not being able to see images. It 
frankly feels rude.


> (* where parsing_model_overview.html would also attempt to explain the 
> visualization in greater detail than the subsequent on-screen text 
> provides)

I'll grant you that some images might benefit from a separate treatment, 
though in practice in fourteen years of browsing the Web I can still count 
the number of images I've seen with such treatment on the fingers of one 
hand, so I don't think it's worth providing a dedicated feature for (all 
such descriptions in fact are just as useful for sighted users and thus <a 
href=""> handles them quite adequately). And that's even with my having 
specifically done extensive research looking for such descriptions. 

However, in this particular case, I really don't see what such a 
description could possibly add.


> The education piece of the spec (and I grant that you have spent time 
> working on that, and that is a good thing) needs to also point out that 
> images are not the appropriate place to be putting large blocks of 
> critical text - something that WAI has been saying for close to a 
> decade.

Sometimes, images are just the best way to convey something. Consider, for 
example, Feynmann diagrams. It would be inappropriate to require the Web 
to convey information only in the least-common-denominator medium simply 
because a minority of users are unable to use the best medium.

Similarly, some situations, such as a blind person uploading their photos 
for their friends to describe to them, have nothing _but_ a visual 
representation. So it would make no sense to suggest using text instead.

Rather, we encourage authors to provide textual alternatives where 
possible.


> Our best understanding at this time is that the majority of daily users of
> screen reading technology generally prefer terse descriptions of images on
> first pass

Really? They prefer terse descriptions of images to just self-contained 
text with no images at all? This seems to run counter to what you've been 
saying up to this point.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Saturday, 25 April 2009 22:17:54 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:39:02 UTC