[html4all] Research? (was RE: some reflections on @alt usage)

[John Foliot]

Charles McCathieNevile wrote:
> All of the stupid advice about
> times when alt can be ommitted needs to be rewritten.

+1

> 
> There are arguments given in the debate that are about blessing bad
> tools, an those specific arguments should be shot down in flames.
> 

...more like blessing bad practice (I note with irony that they no longer
call it paving cowpaths, although the concept remains the same).

> The spec should point out that not having alt is wrong.

Agreed (perhaps *very* wrong?)

> But having
> some default value that is meaningless, or that collides with a
> meaningful value like alt="" is even more wrong since it not only
> breaks the user experience like no alt does, but it also becomes
> harder to test and breaks existing rules, advice, and tools.

Agreed (painful to admit Chaals, but, agreed).  Ditto for shoveling in lousy
alt text just to shut up the validator.

> 
> If we get these issues fixed in the spec then we are ready to start
> addressing the actual question of whether missing alt should be a
> validation error.

It should be (more below).

Actually, I think what is more important is the *consequences* of inserting
visual only content into the information stream without providing the
"alternative".  In my previous posting, I echoed thoughts that Steven had
suggested in an earlier thread (never claimed they were my original thoughts
Steven <grin>): give the authors plenty of different ways - entrenched into
the spec - to ensure that a textual alternative is present, but at the same
time I proposed to up the ante for failing to do /anything/.  My proposal
was drastic to the extreme, yet nobody has commented on this. (Am I really
that far out on the ledge with that one?)

Chaals, I wish we could deal with this issue in the linear fashion you
propose, but this apparently clashes with the "all-in" approach that Ian and
Co. are taking - they want whole solutions now, not partial proposals with
dangling action items: hell you re-opened the whole @alt issue on the
technicality that Ian had attempted to close it even though there was an
official open action item...

> 
> Optional alt is about a belief that validation is more important to
> people (in particular tool developers) than accessibility. If that
> turns out to be true, then it makes sense. If that turns out to be
> false, then it doesn't. But we have to have some research - both
> sides of the argument are currently based on gut feeling and
> instinct. There is a real issue here, and there is very little real
> information being provided to settle the question rationally.

I believe that this is only partially true.  Clearly, to the editor and his
cohorts, validation (conformance) *is* important; and isn't the whole point
of standards is that to do things right, to *ensure* inter-operability,
adherence to said standards is a given?  I think we would be hard pressed
today to find an engineer of any stripe to argue against this concept.

Accessibility advocates want/need the same kind of entrenched requirements,
else the demands/requirements get pushed down a notch: it really is that
simple.  The research you ask for exists already: the millions (trillions?)
of inaccessible web pages in the wild today simply because WCAG 1 is a W3C
Guideline and not a W3C Recommendation.  If we are to advance even the
simplest of needs/requirements beyond where we are today, we need more stick
- full stop.  This might *sound* like gut feeling, but I cannot find any
examples on the web today to counter this feeling, and the fact that we need
to fight **this hard** just to ensure that visual only content is properly
accounted for by authors, simply highlights the fact that my gut feeling is
inherently true.  But I made myself a promise to calm down and reign in the
rhetoric... I believe the concept was winnable battles?

> 
> So my preferred approach is to fix the rest of the rubbish in the
> spec, and do some research on the real issue, and *then* talk about
> making a decision...

What further research needs to be done, exactly?  Way back when, one of the
lofty goals of html4all was to actually get this kind of stuff done, but
that "stuff" needs to be fully articulated and defined.  You are one of the
wise sages here, so give me/us some direction on what kind of research you
believe is required to settle this round-robin and get actual traction
happening.  I will jump in whole-hog, hopefully others as well, and we can
also tap our existing contacts and reach out: there are groups such as GAWDS
that seem to be almost ignorant of what is going on here - ready troops who
would probably jump in "...just because".  What about Fundación Sidar?
Chaals, you still have inroads there, correct?  Web Standards Group?  A
request for help via "AListApart"?  Rounding up troops should not be that
hard (says John, fully realizing that it may be a bit naive to say so, but
I've stuck my neck out before...)

> 
> cheers
> 
> Chaals