W3C home > Mailing lists > Public > public-html@w3.org > April 2008

Re: Request for review of alt and alt value for authoring or publishing tools

From: Leif Halvard Silli <lhs@malform.no>
Date: Wed, 16 Apr 2008 01:16:36 +0200
Message-ID: <48053754.3010408@malform.no>
To: Ian Hickson <ian@hixie.ch>
CC: John Foliot <foliot@wats.ca>, "'Tomas Caspers'" <tomas@tomascaspers.de>, wai-xtech@w3.org, wai-liaison@w3.org, public-html@w3.org, "'HTML4All'" <list@html4all.org>

Ian Hickson 08-04-15 23.14:     
> On Tue, 15 Apr 2008, John Foliot wrote:
> > Ian Hickson wrote:
>   
> > But in this particular case, the spec is excusing a key player (the 
> > authoring tool/web-app) from it's role in ensuring that the playing 
> > field remain level.  You are saying "we can't come up with a solution, 
> > so AT needs to do so" while at the same time giving AT *absolutely* 
> > nothing to work with.
>
> The server has nothing to work wither either. None of the players here 
> have anything to work with. It sucks, but that doesn't make information 
> magically appear out of nowhere.
>   

There ought to be several things to work with on the page you provided 
below: The image caption, the name of the user, date/time, title of the 
photostream.

> > > We can get better accessibility by letting user agents compete on best 
> > > handling of these images than we can by letting servers, who have near 
> > > zero motivation to address this issue, try to come up with some 
> > > half-baked solution.
> > 
> > But if the servers *must* provide part of the solution (to be 
> > "conformant" servers) you have given them the motivation.
>
> Well, we can test that now, can't we? Since HTML4 require the alt="" 
>   

HTML 5 gives many examples. That is an improvement. It is not only 
demanding that matters, but showing the way as well.

> attribute, your argument is that servers have the motivation to include 
> alternative text on images for which the user has not included any 
> alternative text.
>
> So let's look at a random image on Flickr:
>
>    http://flickr.com/photos/18356286@N00/854359279/
>
> What's the alternative text on the critical image?:
>
>    <div id="photoImgDiv854359279" style="width:502px" class="photoImgDiv">
>    <img src="http://farm2.static.flickr.com/1378/854359279_73592d7334.jpg?v=0" 
>    alt="" width="500" height="375" onload="show_notes_initially();" 
>    class="reflect"></div>
>
> The empty string! That's the single _worst_ value you can give in this 
> case. What we learn from this is that requiring alt="" attributes on 
> images ends up motivating the servers _to lie_. They say the image isn't 
> important, that it's decorative, when in fact it is the single most 
> important piece of information on the page.
>
>   

20 of 26 images all together have alt text and are just site stuffing. 
(And 4 of the <A> elements also have alt attributes  ...). The 6 other 
images have emty alt. At least half of those 6 belong to the user/reader 
generated content/stream.

So, there is a some kind of pattern there. If thoe images could have 
meaningful text, it would matter.

> > > Reserved values are just syntactic variants on omitting the attribute. 
> > > There is no practical difference. (Well, other than reserved values 
> > > being significantly less usable in today's UAs, and omitting the 
> > > alt="" attribute being cleaner, which is why the spec says to omit the 
> > > attribute instead of inventing some new reserved value.)
> > 
> > Yes and no.  Reserved values can be programmatically assigned whatever 
> > values/uses a user-agent needs or wants.  By using a reserved value, AT, 
> > all AT not just a particular flavor or brand of AT, can parse the value 
> > and say "oh, one of those... I do this with those" consistently. While 
> > there is a weak semantic value to a reserved value, there is *some* 
> > value, whereas the vacuum of not having any alt value is just that, a 
> > vacuum, and asks essentially for a guess, without providing *ANY* clues.  
> > Visual users can see the photo, non-visual users are discriminated 
> > against by being handed nothing.
>
> This is incorrect. There is absolutely no practical semantic difference 
> from the UA's point of view between an omitted alt="" attribute when 
> omitting hte attribute is defined to mean "the image is critical but has 
> no content" and a special reserved value which is defined to mean "the 
> image is critical but has no content". It is merely a syntactic detail.
>   

I agree that there is no practical semantic difference then. And for the 
sake of semanticness, it could be possible to define several equally 
bad/good solutions. (Some migh prefer no alt, som an alt with a reserved 
word, others would prefer enumeration.)

But there might be a practical user experience difference, depending on 
what the user agent is prepared to make something out of.
-- 
leif halvard silli
Received on Tuesday, 15 April 2008 23:18:09 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:54 UTC