W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > January to March 1998

Re: ALT text survey

From: Colin F Reynolds <colin@the-net-effect.com>
Date: Wed, 11 Feb 1998 01:36:49 +0000
Message-ID: <T0rsByDxCQ40EwoW@the-net-effect.com>
To: w3c-wai-gl@w3.org
Cc: "Alan J. Flavell" <flavell@a5.ph.gla.ac.uk>
[cc: Alan Flavell - since I don't know whether he's on this list]

In article <F9AE637AED42D01187B400A0C913772E8F987B@mailsrvd.ho.att.com>,
"Kasday, Leonard" <kasday@att.com> writes
>       Colin,
>Thanks for the feedback on the standards.   My comments preceded
>by LRK::
>>When this page is viewed with the latest Netscape Navigator
>>browser (4.04/Win95/1024*768 resolution) (with image loading
>>turned off) _no_ ALT text is displayed on my system[1] for
>>_any_ of these images, demonstrating the futility of such
>>precise instructions when authoring content for the WWW.
>LRK:: this is intresting, but surely you don't mean that no alt text is
>shown anywere.  There are some large images, e.g. 145x170.  Which images
>don't show the text?   .  

No, that's exactly what I _do_ mean - on my system NONE of the images
shows _any_ ALT text (except as tooltips!), which is why I said that
giving pixel-perfect instructions is pointless.

>> In the "Recommended Guidelines for New Pages" section 3, the
>use of such
>> >ALT text such as "bullet" for bullet images is recommended; this is
>> >something which I (and others) feel is totally inappropriate. Quoting
>> >from Alan Flavell's study[2] of the use of ALT in IMG:
>> ><blockquote>
>> > So we get to read (or blind readers get to hear): 
>> >        Large Yellow Bullet Introduction
>> >        Large Yellow Bullet The Problem
>> >              Small Red Bullet Historical Analysis
>> >              Small Red Bullet Current Situation
>> >        Large Yellow Bullet The Solution
>> ></blockquote>
>       LRK:: This is not what the AT&T alt text standards recommend.
>For bullet, they simply say to use one word, "bullet",
>except, to quote
>the standards

Which standards?

>       "Include color or description of bullet if it has meaning, e.g.
>if red bullets imply higher importance than blue bullets. "

With respect, you're nit-picking. Substituting "red bullet" for "Large
Yellow Bullet " and "blue bullet" for "Small Red Bullet ", that gives us

      red bulletIntroduction
      red bulletThe Problem
            blue bulletHistorical Analysis
            blue bulletCurrent Situation
      red bulletThe Solution

Hmmm... looks the same to me, if not worse, because of the lack of a
space separator after "bullet" in the ALT content.

In any case, why do you want to use coloured bullets to indicate
importance? The content _already has_ this information, via the nested
list. The browser _should_ be able to grok this and render the
information accordingly.

>       > Firstly; just 11 responses is far too few upon which to base
>       > meaningful conclusions
>LRK:: Although it's a smaller number of responses than I would have
>liked, they were accommpaniend by explanations that made them quite
>compelling to me.   When you say that "this is something which I (and
>others) feel is totally inappropriate"  how many people (you plus
>others) does this represent and how many are blind?

I don't know of any definitive studies, but there have been estimates
bandied around for some time that as many as 30% of users of graphical
browsers use them with image auto-loading turned off. Don't you think
that it's necessary to cater for the needs of these users?

As to how many are blind - that's irrelevant. Blind users can best be
served by development of an appropriate browser which renders the
content accordingly for their special needs: but such a browser is
unlikely to be developed until we have an unambiguous definition of the
_meaning_ of each attribute of each and every element.

ALT is just a very good place to start.

>     At most, I think
>we have to recognize that there are differences of opinion here among
>users that we need to consider.     We can't say that anything is
>"totally inappropriate" if thats what some people prefer and it does no
>large harm to others.

But that's just the point: catering exclusively for the needs of one
minority (eg blind people) disenfranchises another group (eg those who
choose to browse with image loading turned off).

Until we have unambiguous descriptions of the meaning of the markup,
differences of opinion are _bound_ to arise - such is the nature of

>If we could get feedback from a wider cross section of blind users that
>would be great.

Yes, it no doubt would be of great benefit to someone developing a
browser for blind people - but it wouldn't help the issue at hand, that
of determining what ALT means.

May I refer you to a posting made by Alan Flavell on w3c-wai-ig, which
sums up the position neatly:

==== begin included text ====

In article <Pine.OSF.3.96.980210225622.3087A-100000@a5.ph.gla.ac.uk>,
"Alan J. Flavell" <flavell@a5.ph.gla.ac.uk> writes
>I don't primarily think of ALT _as_ it's represented in any particular
>browser, but _for_ its stated purpose.  And its purpose is to serve as
>alternative text, for presentation when the image is - for whatever 
>reason - not being presented.
>"Alternative text".  Let's concentrate on that.
>Whatever you may say about what's going in behind the scenes, there's a
>need for alternative text, and this is it.  If you want to "property
>identify the object" for some other good reason, I'd respectfully suggest
>you look for another attribute.  This one's assigned, in my opinion.  Just
>occasionally, the appropriate alternative text might identify the object,
>describe the object, explain the object, name the object, whatever.  Its
>purpose is none of these things: its _purpose_ is to supply alternative
>text, for use when the object is not being presented.
>It's a compromise, for sure, but this is surely what we should be
>aiming at, no?  TITLE and LONGDESC (for IMG) are other handles we may
>turn, in their respective ways.
>It would be a complete mistake to let our choice be dominated by what one
>or other browser happens to be doing today.  Let's concentrate on the
>functional requirement, in terms of user needs, and then we can evaluate
>how well the browsers are fulfilling that requirement.  Not vice versa. 
==== end included text ====
Colin Reynolds, The Net Effect (World Wide) Ltd
Tel: +44 (0)1246 450 901
Fax: +44 (0)1246 450 902
Received on Tuesday, 10 February 1998 22:14:48 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:26 UTC