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

Re: crystal reports as html

From: Charles McCathieNevile <charles@sidar.org>
Date: Sun, 26 Jan 2003 12:06:57 +1100
Cc: WAI-IG <w3c-wai-ig@w3.org>
To: Joe Clark <joeclark@joeclark.org>
Message-Id: <774A0FCE-30CA-11D7-A311-000A95678F24@sidar.org>

Joe,

as far as I can tell, your suggestion here is to throw up your hands 
and say "it might be too hard". I trust i am misreading something and 
that you really do expect people to interpret data.

Some data (the kind that generates very simple bar charts, for example) 
can be readily transformed into text by a process parallel to the way 
it is transformed into graphs. (I presume you are aware that most of 
the graphs are created automatically, not hand drawn.) An extremely 
simplistic approach is to use the tablin tool originally produced by 
Daniel Dardailler when he was technical director of WAI, but there are 
a number of companies who produce, and advertise on this list, tools 
for rendering tables of data as a text stream. It is also possible, if 
you assume that everybody's software works with standards, to give 
people the original tables of data.

I agree with you that the reason we draw graphs is that many people 
find it hard to understand tabulated numerical data (experience 
suggests that many people don't really understand the graphs either, 
something that advertisers, journalists, lawyers, and others in the 
business of promoting a particular interpretation often rely on). It is 
a big leap (and one that defies logic) to suggest that we should 
therefore not present the raw data - particularly as certain types of 
disability make it impossible to read a graph.

Your own book suggests a simplistic approach to the problem, (see 
"rating stars" in chapter 6 "the image problem") which is very 
effective for the simple cases. I have seen data usefully presented in 
this way - while the data itself (and hence the resulting pages) is 
confidential information, I will ask if the code used to generate them 
can be made public.

Finally, as Joe is already aware, I am not working for WAI any more and 
don't speak on behalf of them in any sense. Given that WAI is really 
just the sum of what its participants can do, if they get it right 
their advice is the collective  best of what participants have to 
offer, but it is also possible that the worst of what we can do is what 
gets collected. Constructive advice about how to improve on WCAG 1 is 
best directed to the working group in the form of comments on working 
drafts of WCAG 2. Reading the latest draft at 
http://www.w3.org/TR/WCAG20 will give you an idea of what the working 
group has done so far, and explains where to send feedback. In this 
instance I hope we can come up with something that is better than the 
example Joe derides, and better than just giving up because it might be 
too hard sometimes.

cheers

Chaals

On Saturday, Jan 25, 2003, at 11:57 Australia/Melbourne, Joe Clark 
wrote:

>
>> If you have access to the original data used to generate the graph
>
> Note: *Data*. Something difficult for humans to understand, which is
> why we draw graphs.
>
>> you
>> should be able to generate a textual explanation of it as well
>
> Here we go again. The WAI has a nasty habit of demanding that Web
> authors create labour-intensive and indeed useless derivative works;
> carries on as if such were even *possible* in the first place; and
> avoids doing the same dirty work itself. WAI has a master-slave
> relationship with authors: You struggle under the boulders so we can
> enjoy the pyramids.
>
> When it comes to long descriptions of charts and graphs, the
> requirement
> <http://www.w3.org/TR/WCAG10-HTML-TECHS/#long-descriptions> gives,
> as its *full and complete* example, only the following:
>
> <quoth>
>    Example.
>    <IMG src="97sales.gif" alt="Sales for 1997"
>         longdesc="sales97.html">
>
>    In sales97.html:
>
>      A chart showing how sales in 1997 progressed. The chart
>      is a bar-chart showing percentage increases in sales
>      by month. Sales in January were up 10% from December 1996,
>      sales in February dropped 3%, ..
>
>    End example.
> </quoth>
>
> Note well the ellipsis. WAI cannot even be bothered to *write out a
> genuine example* in *all* the detail they expect the rest of the
> world to provide. In part this is because WCAG is so tremendously
> poorly written.
>
> It's a fatuous and irrelevant idea anyway. To quote myself (Chapter
> 6, page 85): "like packing, unpacking, and repacking a suitcase,
> once some numbers are rendered graphically, it is not always easy or
> straightforward to re-render that graphic in numbers (and words)."
>
> I note that Charles rather revealingly uses the term "generate":
> "you should be able to generate a textual explanation." Like it's
> produced by a machine. Like it's even possible half the time. Also,
> "a textual explanation" is a clumsy, euphemistic way to demand that
> page authors *write coherent prose*, an ability WAI lacks. You know
> you're in trouble if you're being told to mount an impossible
> campaign, your leader shoulders none of the work, and your leader
> can't even be bothered to tell it to you straight.
>
> WAI members had at the time and possibly still have an incomplete,
> casual, anecdotal knowledge of the true nature of charts and graphs.
> (That's equally true for the true nature of *writing*, but we'll get
> to that another day.)

This is clearly an allegation without substantiating evidence.

> If it were genuinely possible to transform many classes of *data*
> into *coherent prose*, we wouldn't use graphs in those cases. Some
> or even most of the time, it *is not possible* to transform data
> into a graph into prose-- *if* you want to retain even significant
> portions of the underlying meaning, let alone the myriad details
> that become visible in graphing. (I would point to scatter charts,
> error bars, outliers, and logarithmic scales as examples of
> *information* that is not clear from the underlying data and resists
> encapsulation in prose.)
>
> I'm not a huge fan of Edward Tufte's books, but here's a challenge
> for the WAI: Take ten examples from each of his volumes and
> "generate a textual explanation" of them.
> <http://www.edwardtufte.com/1176313877/tufte/books_vdqi>
>
> -- 
>
>   Joe Clark  |  joeclark@joeclark.org
>   Author, _Building Accessible Websites_
>   <http://joeclark.org/access/> | <http://joeclark.org/book/>
>
>
--
Charles McCathieNevile           charles@sidar.org
Fundación SIDAR                       http://www.sidar.org
Received on Saturday, 25 January 2003 20:07:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:08 GMT