W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2003

Re: [320] Ability to be expressed in words

From: Joe Clark <joeclark@joeclark.org>
Date: Tue, 26 Aug 2003 12:06:31 -0400
Message-Id: <a06001fadbb7132df704b@[]>
To: WAI-IG <w3c-wai-ig@w3.org>

I certainly appreciate Al's efforts here (he can always be relied on 
to give a solid theoretical basis), but I don't think the proposed 
WCAG guideline deals with the same thing he's talking about.

>The right goal for our effort, as Tom has suggested, is "what tip of 
>the information iceberg is most important to get articulated 
>verbally, and how can we get authors over the hump to articulate 

Unfortunately, that formulation assumes that a diagram or 
illustration (the canonical example that WCAG continues to flub) has 
one or two easy-to-summarize main points. A *gist*, you might say.

There are lots of those, and I cover them in my book. But they are 
not the norm, and the entire reason we draw diagrams and 
illustrations is to make relationships that are hard to express in 
words visible and comprehensible. Through the diagram or 
illustration, the entire iceberg is available. It simply is not 
possible to reverse-engineer that iceberg into a simple "tip" that 
can be plunked into an alt text.

Some things *cannot be expressed in words*. That's why words are not 
the only form of communication we use. It follows-- pay attention, 
Chaaalz-- that there aren't "alternatives" or "equivalents" for 

>If we can communicate the right questions to the authors, they will 
>be able to articulate the answers.

No, that perpetuates the fallacy. Some or indeed most diagrams and 
illustrations are permanently and unchangeably resistant to summation 
in words. You can't rephrase the question because the question cannot 
be asked.

>The piece of this job that looks doable in the near term is to coach 
>the author through a bit of scene modeling and get them to 
>articulate what is in their scene, some principal properties of each 
>such thing, and principal relationships among these things. Then we 
>will be getting somewhere.
>This will create a guide to their scene in an 
>entity/relationship/attribute graph with navigable relationships and 
>speakable attributes.

People aren't going to put a full workday into modeling, in some 
as-yet-specified language, every relationship that is apparent in an 
illustration. Nobody's gonna write some kind of script for a screen 
reader (similar to Jaws configuration files) to enable a single 
illustration to be read. Even then the experience won't merely fail 
to be the same, it won't even be adequate. *Some things can't be 
summed up in words*.

>[The URCC primitives: "What's there?" "What can I do?" are still the 
>core of what we have to lure out of authors. Imagine yourself 
>answering these questions over the phone...]

I've used that model myself in other contexts. But it doesn't work 
here. Many or even most diagrams and illustrations (to continue with 
the canonical WCAG example) are too complex for that. Essentially, 
WCAG wants the entirety of technical and scientific illustration 
reduced to TV-news soundbites.

>Talking about creating a GIS-like infiltration of the scene 
>presented on the web screen helps the content designer/developer 
>think in graph terms and not be forced to take the harder conceptual 
>leap to a linear narrative.

I see what you're saying (not everything need be turned into lengthy 
prose to be "expressed in words"), but even that amount of metadata 
simply *will not be done* by authors. They've got better things to 
do, and some will realize the futility of equating illustration with 

>Anything we do to lower the potential barrier to getting over the 
>belief that it's too hard is golden, is critical to our success.

WCAG WG and WAI need to accept that "express[ing] in words" isn't 
merely hard some of the time, it's impossible.

*Some things can't be made accessible to everyone*. That's why, for 
example, we have undue-hardship exemptions in human-rights 

>  Presented with a tree-view presentation of the objects they have 
>identified in their scene, an author can readily from this view 
>create a sensible tour, a linear reading order.

There isn't a "linear reading order" inherent in everything. It's a 
bit late in the day to suggest that the Web-- a non-linear medium-- 
must be reduced to a straight line. "Linear reading order" is another 
of those WAI shibboleths that needs to be de-emphasized.

>  >On Thursday, August 21, 2003, at 10:55 AM, Joe Clark wrote:



     Joe Clark | joeclark@joeclark.org
     Accessibility <http://joeclark.org/access/>
     Expect criticism if you top-post
Received on Tuesday, 26 August 2003 12:10:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 13 October 2015 16:21:25 UTC