Unready and social engineering Re: [html4all] several messages about alt

On Mon, 14 Apr 2008 02:44:02 +0200, Dannii <curiousdannii@gmail.com> wrote:

> On Mon, Apr 14, 2008 at 1:33 AM, Leif Halvard Silli <lhs@malform.no>  
> wrote:
>> Those are different evaluation axes. But experience has shown us that  
>> the only validation that authors care about is the general CSS and HTML  
>> stamps.

Hmmm. Experience shows that there are people who care about other stamps,  
including accessibility ones. It also shows comprehensively that all of  
these together are still, with teh Web almost two decades old, things that  
only a small minority of developers care about.

>> Therefore, we must (continue to) incorporate social consciousness into  
>> the general stamping tools.

Henri and others have made it clear that they do not think the necessity  
for this is clearly established. So at the very least there is no  
consensus on this point.

Personally I look at the question slightly differently. The HTML  
specification determines what needs to be done to make a document  
interoperable. For some known classes of users, interoperability relies on  
(and for some more classes is greatly improved by) being able to strip the  
document down to a non-graphic interpretation. This should be clarified by  
the specification.

How that plays out in terms of what "validity" means is still an  
unanswered question - at least in the sense that we still appear to have  
violent disagreement in the working group on the issue (far more  
disagreement than with the principle that I stated above about non-graphic  
representation being necessary to interoperability, for example).

> For an unready stamp to be successful I think all of the following would
> need to be ensured. Can you provide any evidence they would be?
> 1. It really would have to be used by only a small number of pages

No, I don't see why this matters. Validating HTML is still a minority  
activity (very minority) and yet is clearly of value, or we wouldn't have  
this discussion.

> 2. The public would have to continue seeing it as undesirable, rather  
> than accepting and even preferring it to the full spec. Even if the
> validators gave warnings or errors those might soon be regarded as flaws
> of the validator... for example, does anyone actually pay attention to
> the CSS validator messages about not providing both fore and background
> colours?

Yes. Furthermore, do you have any evidence that people actively regard  
these warnings as bad?

The fact that people often don't do what they consider to be "the right  
thing" accords well with everything we know about people, and the fact  
that sometimes they attempt to rationalise that by claiming that somehow  
"the right thing" isn't actually right does too.

So the question boils down to evidence that the stamp is seen as a  
desirable thing overall. Anecdotally, validation of HTML is seen as a  
desirable thing, but is generally held as secondary in priority to pages  
actually working in browsers. For an example, see Ben Buchanan's  
discussion [1] of getting part of "The Australian" (a major Australian  
news site) to validate - and note that almost all of the site still  
doesn't, since it is more important that it actually work, and that is  
defined differently by the people responsible for the site.

> 3. People would actually need to work actively to fix pages with the  
> stamp.

I don't think this is the critical thing, I think that it is more  
important that what the stamp conveys is seen as something worth working  

Which means you would expect an increase over time in the proportion of  
content that merits the stamp. The alt case may be instructive here. A  
decade ago, when WAI was first working, use of the alt attribute was far  
lower and far worse than it is today. The common response to the statement  
that things needed an alt attribute, then as now, was that this was  
unrealistic, or bad for [insert wierd edge case here], or unnecessary for  
[insert current draft's email/document case here - and discussion of it].  
In the intervening decade, despite ongoing discussion about whether it is  
necessary, despite many development tool chains still making it very hard  
to achieve, and despite the fact that it can often seem like an acitvity  
with little real return, my observation is that the prevalence of alt  
attributes, and the quality of use, has increased massively. Again, this  
is anecdotal, based largely on the sites I use everyday or every month,  
but it really is a significant level of change. Almost none of those sites  
actually claim conformance to accessibility guidelines (and nor should  
they since the generally still have major problems) but they have all made  
substantial progress towards being able to do so. Additionally, the  
ability to provide alt has become far more widespread - although there are  
still glaring examples of failure in this area there are also far more  
tools that have improved their "level of conformance" to ATAG [1] and  
similar requirements.

> 4. That the stamp wouldn't be used in more cases than intended. Yes it's
> intended for CMS', but what's to stop it being used on any pages where  
> the author is too lazy to add alt attributes?

Nothing - so we should be realistic about the use cases for any stamp, and  
reckon on the overall cost/benefit from this. Ian asked elsewhere what  
would be the benefit in making a private email between himself and his  
partner non-conforming, but equally it could be asked what possible  
benefit is tehre to them in knowing that their private email *is*  
conforming? In either case I suspect the answer is "none whatesoever", so  
looking at the cost becomes worthwhile...

As a thought exercise, then, if we had such a stamp, what would be the  
drawback of extending its use from CMS' to any lazy author? Personally, I  
see none - and actually some benefits in doing so. I find it very  
difficult to understand why the current draft is prepared to allow some  
classes of tool to claim conformance while being second-rate. But this  
goes to the fact that a lot of the validation discussion is really about  
social engineering, and howto achieve an outcome that is desirable for  
technical reasons but cannot simply be specified into being...



Charles McCathieNevile  Opera Software, Standards Group
     je parle français -- hablo español -- jeg lærer norsk
http://my.opera.com/chaals   Try Opera 9.5: http://snapshot.opera.com

Received on Monday, 14 April 2008 06:50:54 UTC