W3C home > Mailing lists > Public > www-archive@w3.org > September 2007

RE: [html4all] the alt attribute debate

From: John Foliot - WATS.ca <foliot@wats.ca>
Date: Sat, 22 Sep 2007 15:58:05 -0700
To: "'Anne van Kesteren'" <annevk@opera.com>
Cc: <www-archive@w3.org>, <list@html4all.org>
Message-ID: <004501c7fd6c$0a361400$04000100@bosshog>

Anne van Kesteren wrote:
> On Sat, 22 Sep 2007 01:51:15 +0200, John Foliot <foliot@wats.ca>
> wrote: 
>> This might be your justification, but it does not address *any*
>> accessibility issues - it only makes your life easier.
> It's a question of alternatives. If it's a given that the author can't
> provide accurate replacement text, what should be put on the <img>?

Well, first, it's not a given.  You have yet to prove to me that it was a
given that you could not supply an alternative text to your image of that
building in Boston.  You simply stated that the effort could not be
justified by you: "...the single constrain I have: not finding it necessary
to provide replacement text for all those images. This would take too much
time for little benefit."  There is no technical justification, only a
personal perspective - *you* did not find it necessary, and *you* felt that
it would take too much time.

> <img alt=""> or simply leave it <img> so user agents can do image
> analyzing. 

Neither is good, but alt="" indicates that the value is null.  How will the
user agent do this analyzing you speak of, and why could it not also analyze
an image with alt=""?

> Or maybe there's some better construct? 

Perhaps, but to date none has been proposed.

> Something that's given is that authors will make it inaccessible if
> conformance requires that. If for conformance the alt attribute is
> required authors will simply do <img alt=""> and be done with it,
> which doesn't really seem to help anyone.

That's a might big "given".  And so, "for conformance sake" an image can
exist without an alt attribute?  You are simply replacing one bad solution
with another bad solution - but your bad solution makes it easier to be "in
conformance".  It's a shell game. You upload your image of the building in
Boston, and you're "done with it".  How does this help anyone besides you?

> I made this point a few
> times, and you seem to on ignoring it 

I am not ignoring it.  I just responded above, and I will again go on record
as stating that neither solution is good; both are wrong, but having a spec
that pardons and condones a state that will *directly* have a negative
impact on accessibility is "wronger". I say directly because the draft spec
allows, pardons and permits pages to have images with no alternative text:
it allows the page to remain in conformance while at the same time
deliberately excluding at least one user-group.

I have seen absolutely nothing that shows how alt="" can be *worse* than
having no alt attribute at all, and we have plenty of "cowpath" evidence
today that shows that current technology prefers alt="" over no alt
attribute.  Yet this has never been acknowledged either.

> assuming WYSIWYG editors will
> solve this. I don't really buy that, they haven't for the past 10
> years and them suddenly making their application harder to use for
> the average end user is unlikely. (Yes, I've heard of one that asks
> users whether it's a content or presentational image and then
> requires alt= in case of a content image.) 
> The point you seem to try to make is that this problem, that we've
> been having for over 10 years, will suddenly solve itself by WYSIWYG
> editors and the like requiring replacement text to be provided.

...and so the answer is to permit bad behaviour, "harmful" behavior, and
allow a document to none-the-less be "conforming"?  This is not a solution,
it's a Get-Out-Of-Jail-Free card

First of all, who exactly has stated that the current poor solution of
alt="" is a problem?  It's less than ideal, no argument, and it does nothing
for accessibility, but problem?  State the problem:

Images without alternative text are inaccessible to non-visual UA's and
users of those UA's.

OK, that's a problem.  How does allowing an image to exist in a page without
an alternative text solve the problem?  Oh, it makes it easier to be in
conformance, it makes it easier on the author who cannot justify the effort
required, and it makes it easier for the Flickr's of the world, but *HOW
DOES IT SOLVE THE PROBLEM*?  The image is still inaccessible to that end

> I
> don't really buy that. There's also evidence that such content
> producers will harm end users solely to make their pages conforming
> to HTML. So given that, what's the best solution for the end users
> given those content providers? 

What is the harm you reference?  I know first hand that currently, images
with alt="" are not voiced by screen reading technology; where-as images
without any alt attribute are generally voiced with the file information
"image: DC27456 dot jaypeg".  Which is less harmful?  Therefore, from what I
know first hand currently the best solution is alt="".

>> How can a group that wants to dismiss LONGDESC because it is flawed
>> at the same time condone authoring tools that are flawed and seek to
>> write a SPEC to forgive those flaws?
> It's not a flaw in authoring tools, they won't solve this problem.
> Once an authoring tool starts to require it I would expect users to
> simply switch to one that doesn't. We're trying to have a realistic
> specification, that doesn't expect authoring tools to solve problems
> which they haven't done in the past 10 years.

A realistic spec?  One that references a lead-to-gold promise of heuristic
image analysis?  Anne, please.  Apples to apples and oranges to oranges.
You state that authoring tools will never evolve to solve the problem, but
some magical future tool will emerge to provide a better result?  Now I'm
the one having a hard time buying in.

>> The spec should insist that inline images (as opposed to CSS supplied
>> "decorative" images) have alternative text, and preferably something
>> that is user friendly - full stop.  The author group needs to stop
>> trying to determine what is user friendly... You can no more decide
>> what works for any given user than you can ascertain what flavor of
>> ice-cream they prefer (or even if they eat ice-cream...)
> And you somehow can?

No, I cannot, and have never presumed nor stated such.  This is why I can
accept that even alt="photo of a building" is better than "image: zero zero
nine dot jaypeg" [http://anne.is.weggeweest.nl/2006/boston/009.jpg] -
because for some users that *might* be sufficient.  I don't know, and do not
want to presume either way.

>> JF (trying *very* hard to remain rational, on topic, and respectful)
> workman", etc.

1)  I used emphasis to point out the issue: that in your piece you stated
that the reason for not providing an alt value was not that it was
technically impossible to do, but rather that in *your opinion* it was too
much effort for too little pay-back. 

Every time somebody strenuously debates with you guys, you claim we are
grandstanding or being argumentative and negative.  The written word is a
difficult medium to express nuance, and so I must use imperfect means of
applying emphasis.  

And so, without all caps:  Anne, the point that *I* am making is that you
used a loophole in the draft spec to avoid adding alternative text - it was
simply easier for you to upload the photo of the building in Boston and move
on - no problem, my page is conformant (problem solved). This is wrong, and
if you can (and did) do it, then multitudes of others after you will do the
same, and accessibility will be the whipping post.  In my opinion, there is
nothing that can justify this - nothing!  That it will continue to happen I
must accept - that it is "permissible" by virtue of an international
standards document I cannot.  And *that* is the crux of the argument.

2)  I am not saying that *you* are a poor workman, but rather countering the
proposed solution that presumes it's a strictly technological issue - and
one that can be solved by technology, by stating an old-country expression
that says that "it is a poor workman who blames his tools":

"Anyone who works in software development, as I do, knows that technological
determinism is a crock.  Technological determinism is the reification of
technology. Because we love our tools, we assign them magical properties,
which simultaneously let us off the hook for our own actions. The tool has
been well analyzed as an extension of the human mind or body, but poorly
analyzed as an excuse. Blaming our technology for what we are is the moral
equivalent of the Twinkie defense."

"It is seldom the case that shoddy workmanship is due to inadequate
equipment or difficult working conditions. It results, for the most part,
from the fact that the workman himself is shoddy and indifferent in his

(This by the way, was an expression I first learned from my British
Grandmother over a quarter century ago.  It was not meant as a slight
against you personally.)

Creating a spec that condones images without alternative text might make
life easier for indifferent content creators, it does *NOTHING* to improve
accessibility - and I have seen nothing from anyone to counter that

Received on Saturday, 22 September 2007 22:58:18 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:43:14 UTC