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

RE: Mandatory and Important

From: Justin James <j_james@mindspring.com>
Date: Thu, 21 Aug 2008 16:00:24 -0400
To: "'Matt Morgan-May'" <mattmay@adobe.com>, "'Laura Carlson'" <laura.lee.carlson@gmail.com>, "'Doug Schepers'" <schepers@w3.org>
Cc: "'Karl Dubost'" <karl@w3.org>, "'Ian Hickson'" <ian@hixie.ch>, "'HTML WG'" <public-html@w3.org>, "'W3C WAI-XTECH'" <wai-xtech@w3.org>, <wai-liaison@w3.org>, "'John Foliot'" <foliot@wats.ca>, "'Gez Lemon'" <gez.lemon@gmail.com>, "'Al Gilman'" <Alfred.S.Gilman@IEEE.org>, <w3c-wai-pf@w3.org>
Message-ID: <065501c903c8$8cb64db0$a622e910$@com>

> -----Original Message-----
> From: Matt Morgan-May [mailto:mattmay@adobe.com]
> Sent: Thursday, August 21, 2008 2:53 PM
> To: Justin James; 'Laura Carlson'; 'Doug Schepers'
> Cc: 'Karl Dubost'; Ian Hickson; 'HTML WG'; 'W3C WAI-XTECH'; wai-
> liaison@w3.org; 'John Foliot'; 'Gez Lemon'; Al Gilman; w3c-wai-
> pf@w3.org
> Subject: Re: Mandatory and Important
> 
> On 8/21/08 7:25 AM, "Justin James" <j_james@mindspring.com> wrote:
> > HTML authoring tools will gladly *not* provide error messages for
> this if user
> > testing deems it too obnoxious.
> 
> First, I have to say that after 3 years working on the Authoring Tool
> Accessibility Guidelines WG, prompting for @alt has never come up as an
> issue with any tool vendor I've spoken with. The only issue is
> awareness
> that this is needed. And ATAG doesn't even require that behavior, if
> the
> user wants to shut it off. I consider this line of reasoning a red
> herring.

Then you need to re-read Laura's message. She seems to be under the
impression that making @alt mandatory will create a huge surge in its usage.
Experience with HTML 4 has shown that this is *clearly* not the case. A
large problem with this, is that authoring tools don't demand @alt. Thus, my
"red herring".

So, if the authoring tools don't demand @alt, how exactly do you think that
making @alt mandatory will affect the majority of HTML documents?

> Second, I really have a problem with the argument that we should demand
> less
> from authoring tool vendors because we don't want to annoy them. Tool
> vendors, like everyone else in this ecosystem, are free to adopt or
> ignore
> whatever they see fit to, and they do. Certainly over the last dozen or
> so
> years, we have little experience with authoring tools (or UAs, or
> authors...) actually implementing any spec 100%, so why should this be
> a
> blocker here?

I think you need to re-read what I wrote. *NOWHERE* did I say that @alt
should be optional in order to not "annoy tool vendors". Indeed, what I
wrote was that tool vendors will not make their tools require @alt,
regardless of whether or not the HTML spec requires it, because this will
annoy *users of the tools*. World of difference. Tool vendors do what their
users want and demand. If the tools are a hassle to use (like endless
prompts for @alt), people will use other tools that are less annoying. So
what is going to happen? Tool vendors will *not* require that their users
provide @alt content, and just silently inject an empty @alt.

So congratulations, we've accomplished precisely nothing for people who need
@alt, all we have done is (at best) managed to get tools to inject a bunch
of empty @alt attributes.

Given the absence of @alt on the Web today, it is clear that mandating @alt
is not the solution to this problem in the slightest.

> And if that _is_ a driving factor, then as a tool vendor, I'm rather
> annoyed
> that Dreamweaver, which generates standards-compliant HTML 4.01 and
> XHTML
> 1.0-1.1 code, will now generate invalid table code in HTML5, since
> @summary
> has gone away, and DW generates it by default, in conformance with WCAG
> and
> ATAG. So, which is it? Do authoring tools already in the market matter,
> or
> not?

Again, you've grossly misread me.

> > @src is required for technical reasons...
> 
> No, you just think it is. I think you'll find that most people on "the
> accessibility side" would argue that @alt is as necessary technically
> as
> @src. And maybe some SemWeb folks would argue that @alt is _more_
> necessary.

OK, I think that's you've managed to cross the border of reason. The
*entire* point of <img> is to... display an image. That requires a source of
data. That means that @src is required for technical resons. @height and
@width are not, since those can be determined from the image itself. @alt is
not required to accomplish what @img needs to do. In fact, in many cases
(such as decorative images, or [shudder] "spacer" images), @alt makes life
miserable not just for sighted users, but those with accessibility needs
too.

> > But I also recognize that making @alt
> > mandatory is more a matter of principle and less of an effective
> measure that
> > will actually accomplish anything.
> 
> Then you've discounted all the people who have actually done @alt
> because
> the validator says to. In every case I'm aware of where an opt-out
> becomes
> an opt-in, performance rates plummet. An extreme example is spam: if it
> were

The people who write code by hand are a minority compared to those using
tools. Those who then run their code through a validator are even less. Look
at the stats on the amount of valid HTML 4 code on the planet. People can't
even get doctype correct, do you honestly think that making @alt mandatory
will lead to a sudden surge in @alt usage?

> opt-in, who'd subscribe? Or voting: the US hovers around 45%.
> Australia,
> with compulsory voting, is around 95%. You might raise the issue of
> quality
> here, but frankly, I'd rather have 95% of images with @alt of mixed
> quality
> and the other 5% invalid, than have 30% of images with @alt of good
> quality
> and allowing the authors of the other 70% not to bother.

And once again , you fall into the exact line of thinking that my message
criticized Laura for.

Why do you (and Laura) seem to believe that the HTML 5 spec carries the same
weight with people as laws? What do you think happens to people who violate
the spec, the police come and arrest them, or fine them? Please stop making
comparisons to actual laws, and please stop thinking that making @alt
mandatory will mean it suddenly gets used.

I'd also suggest that you take a look a lot of the messages I've sent in the
past on this topic. I have always been a big supporter of the accessibility
cause. My record in public is extremely pro-accessibility and pro-usability.
Indeed, many of my objections to things in HTML 5 are based around the idea
that it promotes development techniques like AJAX that are extremely
accessibility unfriendly. While I may not be a special needs user myself, I
am very much in favor of making the Web as accessible as possible for
everyone.

But I am also a realist who has been around the block quite a few times, and
I'm telling you, just because @alt is mandatory does not mean that
accessibility will be there, it just means that you are going to get a bunch
of empty @alt's all over the place, at best.

And this is *precisely* why I support Karl's suggestion that we make @alt
mandatory (despite the fact that I feel that it is a meaningless gesture),
bring @role into force across the board, and refer HTML authors to the WCAG
for accessibility guidance. This allows the HTML 5 spec to gracefully handle
HTML authors who don't care about accessibility without beating them over
the head with it (which will in turn cause ANOTHER decade of spec-friendly
people groaning about all of the non-compliant code out there), while at the
same time, providing people who care about accessibility with a full toolkit
to provide as much or as little accessibility as they need or want to.

I am really not sure how much more you can truly expect from the HTML 5
spec.

J.Ja
Received on Thursday, 21 August 2008 20:01:39 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:22 GMT