W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2002

Re: Creating accessible tables for layout and data: alt attributes

From: Patrick Burke <burke@ucla.edu>
Date: Thu, 31 Jan 2002 09:59:07 -0800
Message-Id: <5.1.0.14.2.20020131093649.01d8cdf0@pop.bol.ucla.edu>
To: w3c-wai-ig@w3.org
I think Kynn has stated the optimum solution pretty well.

At 08:35 AM 1/31/2002, kynn-eda@idyllmtn.com wrote:
>Joel wrote:
> > However, I feel user is the one who should decide whether to read or ignore
> > the information.  If the author can add simple alt text to an image to
> > convey what's in the image, then user has the option whether to 
> interpret it
> > or not.
>
>Right, but it's also possible to overdo things. It's possible to add
>complex alt text to all images which could ultimately distract from
>the purpose of the page.
>
This is the problem I had with the previous example that was discussed: a 
background image with the purpose of highlighting the importance of the 
main text. If you have a 200-word ALT that tells me all about the image & 
how mindbogglingly important the upcoming text is, then you have 
effectively hindered me (as a screen reader user) from ever getting to the 
important text. So the ALT defeats the purpose of the image.


>[...]


>For example, let's say that I'm writing a column for an online web
>zine. Along with the column, there's a picture of me. The question
>now is how do you label the picture of me?
>
>Is it important that I'm a white man? That wear eyeglasses? That I'm
>slightly overweight? That I have a goatee? That I have blue eyes and
>blond hair? That I am 6'2"? That I am wearing a suit or a sweater or
>a tie or no tie? Does the color of my tie matter? What if I'm wearing
>a pin which can't be made out clearly but it's obvious that I'm
>wearing one? What about my age? What about my probable ethnic
>background and possibly even social class? What about the lighting
>where I am, and the background?
>
>This is all information which _could_ be conveyed in the visual
>image. The question then is "what needs to be represented, either in
>alt text or in a long description?"
Isn't that exactly the purpose of LONGDESC?: The user has the option of 
finding out the details about the image, & can bail out & return to the 
main document at any time. So there is no downside to going hog-wild & 
giving the date of a photograph, names of participants (with ethnic & class 
status etc if desired). I agree that the author is the one to decide what 
goes here, but I find no downside to a long LONGDESC (other than gobbling 
of server space & author's writing time).

To return to the background image example. If I as a blind Web designer 
come across a LONGDESC describing the way the background highlights the 
main text of the page, & stating that the image is not 
copyright-restricted, then I might decide to use it on my own page. So the 
same text that would be detrimental to access as an ALT could be very 
valuable as a LONGDESC. Seems obvious to me but I guess it's worth saying.

Patrick
Received on Thursday, 31 January 2002 12:58:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:00 GMT